AI Email Enhancer

AI Email Enhancer — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

  • Cups (app)

    Cups (app)

    Cups (stylized as CUPS) was a mobile app launched in New York City in April 2014. It was a mobile payment and discovery platform for independent coffee shops nearby. The app was active in more than 400 cafes in New York, San Francisco, Philadelphia, Nashville, Minneapolis and Saint Paul, and other U.S. cities. == History == Cups was founded in Israel in 2012 by Gilad Rotem and four other co-founders, who were all high school friends. The company ran a limited beta pilot in Tel Aviv and Jerusalem, featuring 80 locations, from September 2012 until September 2014. Customers received all-you-can-drink coffee at certain coffee shops in Tel Aviv for approximately $45 a month. In October 2013, the founders relocated to New York. Cups participated in the Entrepreneur's Roundtable Accelerator program and went live in New York in 2014, initially working with 50 small coffee shops in Manhattan and Brooklyn. In early 2016, the company launched 30 locations in Philadelphia in February, followed by 40 more locations in San Francisco in March. == Functionality == The Cups app gave the user a list of the nearest participating coffee shops to their current location. The app user can order a drink using the app and pay the cashier with their phone. The cashier would enter a code that entered the purchase into the app's system. The app also allowed for onboard tipping and food purchases. The company reimbursed the coffee shop and kept a portion of their sales. In early 2016, the Cups Café Network was launched, using bulk purchasing power to land discounts with service providers which would normally be reserved for larger chains. In this way, the company aimed to help its café partners compete with the larger coffee chains.

    Read more →
  • Wearable computer

    Wearable computer

    A wearable computer, also known as a body-borne computer or wearable, is a computing device worn on the body. The definition of 'wearable computer' may be narrow or broad, extending to smartphones or even ordinary wristwatches. Wearables may be for general use, in which case they are just a particularly small example of mobile computing. Alternatively, they may be for specialized purposes such as fitness trackers. They may incorporate special sensors such as accelerometers, heart rate monitors, or on the more advanced side, electrocardiogram (ECG) and blood oxygen saturation (SpO2) monitors. Under the definition of wearable computers, we also include novel user interfaces such as Google Glass, an optical head-mounted display controlled by gestures. It may be that specialized wearables will evolve into general all-in-one devices, as happened with the convergence of PDAs and mobile phones into smartphones. Wearables are typically worn on the wrist (e.g. fitness trackers), hung from the neck (like a necklace), strapped to the arm or leg (electronic tagging), or on the head (as glasses or a helmet), though some have been located elsewhere (e.g. on a finger or in a shoe). Devices carried in a pocket or bag – such as smartphones and before them, pocket calculators and PDAs, may or may not be regarded as 'worn'. Wearable computers have various technical issues common to other mobile computing, such as batteries, heat dissipation, software architectures, wireless and personal area networks, and data management. Many wearable computers are active all the time, e.g. processing or recording data continuously. == Applications == Wearable computers are not only limited to computers such as fitness trackers that are worn on wrists; they also include wearables such as heart pacemakers and other prosthetics. They are used most often in research that focuses on behavioral modeling, health monitoring systems, IT and media development, where the person wearing the computer actually moves or is otherwise engaged with his or her surroundings. Wearable computers have been used for the following: general-purpose computing (e.g. smartphones and smartwatches) sensory integration, e.g. to help people see better or understand the world better (whether in task-specific applications like camera-based welding helmets or for everyday use like Google Glass) behavioral modeling health care monitoring systems service management electronic textiles and fashion design, e.g. Microsoft's 2011 prototype "The Printing Dress". Wearable computing is the subject of active research, especially the form-factor and location on the body, with areas of study including user interface design, augmented reality, and pattern recognition. The use of wearables for specific applications, for compensating disabilities or supporting elderly people steadily increases. == Operating systems == The dominant operating systems for wearable computing are: FreeRTOS is a real-time operating system kernel for embedded devices; most of the Smartbands that are currently available in the market are based on FreeRTOS, which include Huawei, Honor, Lenovo, realme, TCL and Xiaomi smartbands. LiteOS is a lightweight open source real-time operating system that is part of Huawei's "1+8+N" Internet of Things solution. Tizen OS from Samsung (there was an announcement in May 2021 that Wear OS and Tizen OS will merge and will be called simply Wear.) watchOS watchOS is a proprietary mobile operating system developed by Apple Inc. to run on the Apple Watch. Wear OS Wear OS (previously known as Android Wear) is a smartwatch operating system developed by Google Inc. == History == Due to the varied definitions of wearable and computer, the first wearable computer could be as early as the first abacus on a necklace, a 16th-century abacus ring, a wristwatch and 'finger-watch' owned by Queen Elizabeth I of England, or the covert timing devices hidden in shoes to cheat at roulette by Thorp and Shannon in the 1960s and 1970s. However, a general-purpose computer is not merely a time-keeping or calculating device, but rather a user-programmable item for arbitrary complex algorithms, interfacing, and data management. By this definition, the wearable computer was invented by Steve Mann, in the late 1970s: Steve Mann, a professor at the University of Toronto, was hailed as the father of the wearable computer and the ISSCC's first virtual panelist, by moderator Woodward Yang of Harvard University (Cambridge Mass.). The development of wearable items has taken several steps of miniaturization from discrete electronics over hybrid designs to fully integrated designs, where just one processor chip, a battery, and some interface conditioning items make the whole unit. === 1500s === Queen Elizabeth I of England received a watch from Robert Dudley in 1571, as a New Year's present; it may have been worn on the forearm rather than the wrist. She also possessed a 'finger-watch' set in a ring, with an alarm that prodded her finger. === 1600s === The Qing dynasty saw the introduction of a fully functional abacus on a ring, which could be used while it was being worn. === 1960s === In 1961, mathematicians Edward O. Thorp and Claude Shannon built some computerized timing devices to help them win a game of roulette. One such timer was concealed in a shoe and another in a pack of cigarettes. Various versions of this apparatus were built in the 1960s and 1970s. Thorp refers to himself as the inventor of the first "wearable computer". In other variations, the system was a concealed cigarette-pack-sized analog computer designed to predict the motion of roulette wheels. A data-taker would use microswitches hidden in his shoes to indicate the speed of the roulette wheel, and the computer would indicate an octant of the roulette wheel to bet on by sending musical tones via radio to a miniature speaker hidden in a collaborator's ear canal. The system was successfully tested in Las Vegas in June 1961, but hardware issues with the speaker wires prevented it from being used beyond test runs. This was not a wearable computer because it could not be re-purposed during use; rather it was an example of task-specific hardware. This work was kept secret until it was first mentioned in Thorp's book Beat the Dealer (revised ed.) in 1966 and later published in detail in 1969. === 1970s === Pocket calculators became mass-market devices in 1970, starting in Japan. Programmable calculators followed in the late 1970s, being somewhat more general-purpose computers. The HP-01 algebraic calculator watch by Hewlett-Packard was released in 1977. A camera-to-tactile vest for the blind, launched by C.C. Collins in 1977, converted images into a 1024-point, ten-inch square tactile grid on a vest. === 1980s === The 1980s saw the rise of more general-purpose wearable computers. In 1981, Steve Mann designed and built a backpack-mounted 6502-based wearable multimedia computer with text, graphics, and multimedia capability, as well as video capability (cameras and other photographic systems). Mann went on to be an early and active researcher in the wearables field, especially known for his 1994 creation of the Wearable Wireless Webcam, the first example of lifelogging. Seiko Epson released the RC-20 Wrist Computer in 1984. It was an early smartwatch, powered by a computer on a chip. In 1989, Reflection Technology marketed the Private Eye head-mounted display, which scans a vertical array of LEDs across the visual field using a vibrating mirror. This display gave rise to several hobbyist and research wearables, including Gerald "Chip" Maguire's IBM/Columbia University Student Electronic Notebook, Doug Platt's Hip-PC, and Carnegie Mellon University's VuMan 1 in 1991. The Student Electronic Notebook consisted of the Private Eye, Toshiba diskless AIX notebook computers (prototypes), a stylus based input system and a virtual keyboard. It used direct-sequence spread spectrum radio links to provide all the usual TCP/IP based services, including NFS mounted file systems and X11, which all ran in the Andrew Project environment. The Hip-PC included an Agenda palmtop used as a chording keyboard attached to the belt and a 1.44 megabyte floppy drive. Later versions incorporated additional equipment from Park Engineering. The system debuted at "The Lap and Palmtop Expo" on 16 April 1991. VuMan 1 was developed as part of a Summer-term course at Carnegie Mellon's Engineering Design Research Center, and was intended for viewing house blueprints. Input was through a three-button unit worn on the belt, and output was through Reflection Tech's Private Eye. The CPU was an 8 MHz 80188 processor with 0.5 MB ROM. === 1990s === In the 1990s PDAs became widely used, and in 1999 were combined with mobile phones in Japan to produce the first mass-market smartphone. In 1993, the Private Eye was used in Thad Starner's wearable, based on Doug Platt's system and built from a kit from Park Enterprises, a Pri

    Read more →
  • Regulation of algorithms

    Regulation of algorithms

    Regulation of algorithms, or algorithmic regulation, is the creation of laws, rules and public sector policies for promotion and regulation of algorithms, particularly in artificial intelligence and machine learning. For the subset of AI algorithms, the term regulation of artificial intelligence is used. The regulatory and policy landscape for artificial intelligence (AI) is an emerging issue in jurisdictions globally, including in the European Union. Regulation of AI is considered necessary to both encourage AI and manage associated risks, but challenging. Another emerging topic is the regulation of blockchain algorithms (Use of the smart contracts must be regulated) and is mentioned along with regulation of AI algorithms. Many countries have enacted regulations of high frequency trades, which is shifting due to technological progress into the realm of AI algorithms. The motivation for regulation of algorithms is the apprehension of losing control over the algorithms, whose impact on human life increases. Multiple countries have already introduced regulations in case of automated credit score calculation—right to explanation is mandatory for those algorithms. For example, The IEEE has begun developing a new standard to explicitly address ethical issues and the values of potential future users. Bias, transparency, and ethics concerns have emerged with respect to the use of algorithms in diverse domains ranging from criminal justice to healthcare—many fear that artificial intelligence could replicate existing social inequalities along race, class, gender, and sexuality lines. == Regulation of artificial intelligence == === Public discussion === In 2016, Joy Buolamwini founded Algorithmic Justice League after a personal experience with biased facial detection software in order to raise awareness of the social implications of artificial intelligence through art and research. In 2017 Elon Musk advocated regulation of algorithms in the context of the existential risk from artificial general intelligence. According to NPR, the Tesla CEO was "clearly not thrilled" to be advocating for government scrutiny that could impact his own industry, but believed the risks of going completely without oversight are too high: "Normally the way regulations are set up is when a bunch of bad things happen, there's a public outcry, and after many years a regulatory agency is set up to regulate that industry. It takes forever. That, in the past, has been bad but not something which represented a fundamental risk to the existence of civilisation." In response, some politicians expressed skepticism about the wisdom of regulating a technology that is still in development. Responding both to Musk and to February 2017 proposals by European Union lawmakers to regulate AI and robotics, Intel CEO Brian Krzanich has argued that artificial intelligence is in its infancy and that it is too early to regulate the technology. Instead of trying to regulate the technology itself, some scholars suggest to rather develop common norms including requirements for the testing and transparency of algorithms, possibly in combination with some form of warranty. One suggestion has been for the development of a global governance board to regulate AI development. In 2020, the European Union published its draft strategy paper for promoting and regulating AI. Algorithmic tacit collusion is a legally dubious antitrust practise committed by means of algorithms, which the courts are not able to prosecute. This danger concerns scientists and regulators in EU, US and beyond. European Commissioner Margrethe Vestager mentioned an early example of algorithmic tacit collusion in her speech on "Algorithms and Collusion" on March 16, 2017, described as follows: "A few years ago, two companies were selling a textbook called The Making of a Fly. One of those sellers used an algorithm which essentially matched its rival’s price. That rival had an algorithm which always set a price 27% higher than the first. The result was that prices kept spiralling upwards, until finally someone noticed what was going on, and adjusted the price manually. By that time, the book was selling – or rather, not selling – for 23 million dollars a copy." In 2018, the Netherlands employed an algorithmic system SyRI (Systeem Risico Indicatie) to detect citizens perceived being high risk for committing welfare fraud, which quietly flagged thousands of people to investigators. This caused a public protest. The district court of Hague shut down SyRI referencing Article 8 of the European Convention on Human Rights (ECHR). In 2020, algorithms assigning exam grades to students in the UK sparked open protest under the banner "Fuck the algorithm." This protest was successful and the grades were taken back. In 2024, the Munich Convention on AI, Data and Human Rights was introduced as part of growing international efforts to regulate artificial intelligence through a human rights lens. Developed through a collaborative drafting process involving scholars from the Technical University of Munich, Stellenbosch University, Ulster University, and KNUST, the initiative calls for an international conversation on a binding treaty to safeguard human rights and the principles enshrined in the UN Charter in the age of AI. === Implementation === AI law and regulations can be divided into three main topics, namely governance of autonomous intelligence systems, responsibility and accountability for the systems, and privacy and safety issues. The development of public sector strategies for management and regulation of AI has been increasingly deemed necessary at the local, national, and international levels and in fields from public service management to law enforcement, the financial sector, robotics, the military, and international law. There are many concerns that there is not enough visibility and monitoring of AI in these sectors. In the United States financial sector, for example, there have been calls for the Consumer Financial Protection Bureau to more closely examine source code and algorithms when conducting audits of financial institutions' non-public data. In the United States, on January 7, 2019, following an Executive Order on 'Maintaining American Leadership in Artificial Intelligence', the White House's Office of Science and Technology Policy released a draft Guidance for Regulation of Artificial Intelligence Applications, which includes ten principles for United States agencies when deciding whether and how to regulate AI. In response, the National Institute of Standards and Technology has released a position paper, the National Security Commission on Artificial Intelligence has published an interim report, and the Defense Innovation Board has issued recommendations on the ethical use of AI. In April 2016, for the first time in more than two decades, the European Parliament adopted a set of comprehensive regulations for the collection, storage, and use of personal information, the General Data Protection Regulation (GDPR)1 (European Union, Parliament and Council 2016). The GDPR's policy on the right of citizens to receive an explanation for algorithmic decisions highlights the pressing importance of human interpretability in algorithm design. In 2016, China published a position paper questioning the adequacy of existing international law to address the eventuality of fully autonomous weapons, becoming the first permanent member of the U.N. Security Council to broach the issue, and leading to proposals for global regulation. In the United States, steering on regulating security-related AI is provided by the National Security Commission on Artificial Intelligence. In 2017, the U.K. Vehicle Technology and Aviation Bill imposes liability on the owner of an uninsured automated vehicle when driving itself and makes provisions for cases where the owner has made "unauthorized alterations" to the vehicle or failed to update its software. Further ethical issues arise when, e.g., a self-driving car swerves to avoid a pedestrian and causes a fatal accident. In 2021, the European Commission proposed the Artificial Intelligence Act. == Algorithm certification == There is a concept of algorithm certification emerging as a method of regulating algorithms. Algorithm certification involves auditing whether the algorithm used during the life cycle 1) conforms to the protocoled requirements (e.g., for correctness, completeness, consistency, and accuracy); 2) satisfies the standards, practices, and conventions; and 3) solves the right problem (e.g., correctly model physical laws), and satisfies the intended use and user needs in the operational environment. == Regulation of blockchain algorithms == Blockchain systems provide transparent and fixed records of transactions and hereby contradict the goal of the European GDPR, which is to give individuals full control of their private data. By implementing the Decree on Development of Digital Economy, Bel

    Read more →
  • Cancer Likelihood in Plasma

    Cancer Likelihood in Plasma

    Cancer Likelihood in Plasma (CLiP) refers to a set of ensemble learning methods for integrating various genomic features useful for the noninvasive detection of early cancers from blood plasma. An application of this technique for early detection of lung cancer (Lung-CLiP) was originally described by Chabon et al. (2020) from the labs of Ash Alizadeh and Max Diehn at Stanford. This method relies on several improvements to cancer personalized profiling by deep sequencing (CAPP-Seq) for analysis of circulating tumor DNA (ctDNA). The CLiP technique integrates multiple distinctive genomic features of a cancer of interest findings within a machine-learning framework for cancer detection. For example, studies have shown that the majority of somatic mutations found in cell-free DNA (cfDNA) are not tumor derived, but instead reflect clonal hematopoeisis (also known as CHIP). Even though CHIP tends to target specific genes, it also involves many generally non-recurrent mutations that can be shed from leukocytes and detected in cfDNA, regardless of whether profiling patients with cancer and healthy adults. However, genuine tumor derived ctDNA mutations can be distinguished from CHIP-derived mutations. This is because unlike tumor-derived mutations, CHIP-derived mutations that are shed from leukocytes into plasma tend to occur on longer cfDNA fragments, and to lack specific mutational signatures such as those associated with tobacco smoking in lung cancer that are also found in tumor derived ctDNA molecules. CLiP integrates these features within hierarchical ensemble machine learning models that consider somatic mutations and copy number alternations, among other features. While the CLiP method is unique in relying exclusively on mutations and copy number alterations, it is related to a variety of other liquid biopsy methods being commercially developed for early cancer detection using ctDNA and proteins (e.g., CancerSEEK / DETECT-A ), cfDNA fragmentation patterns (e.g., DELFI), and DNA methylation (e.g., cfMeDIP-Seq, Grail). While the CLiP method has not yet been broadly applied for population-based cancer screening, it has been shown to distinguish discriminate early-stage lung cancers from risk-matched controls across multiple cohorts of patients enrolled across the US.

    Read more →
  • Abillion

    Abillion

    abillion was a mobile application helping users to find vegan and sustainable products. The platform allowed users to review plant-based, cruelty-free and sustainable products, while donating between 0.10 and $1 to nonprofit organisations for each review written. As of May 2023, the company claimed to have donated over $2.8M to various nonprofit organisations including Sea Shepherd and Mercy for Animals. The main objective of the company was to reach the number of one billion people following a vegan diet and lifestyle by 2030. == History == The American entrepreneur Vikas Garg founded the company in Singapore and the app was officially launched in May 2018. The start-up was first named 'abillionveg' and changed its name in 2020 to shorten it to 'abillion'. In 2019, the company raised $3M in its first round of funding (pre-Series A). In 2021, it raised $10M in its Series A funding. In February 2023, the company announced the launch of a community investment round, using the crowdfunding platform Wefunder, which reached a total of $500 000. In May 2023, it celebrated its 5th anniversary and reaching 1M downloads. In March 2026, the company announced that they would be closing down by the end of the month. == Awards == Using data from the reviews published by its users, abillion was awarding the most liked vegan products and brands. In May 2023, the company published a world Top 10 Best Plant Based Burgers, among the winning brands were Beyond Meat, NotCo and Sojasun.

    Read more →
  • Algorithmic Puzzles

    Algorithmic Puzzles

    Algorithmic Puzzles is a book of puzzles based on computational thinking. It was written by computer scientists Anany and Maria Levitin, and published in 2011 by Oxford University Press. == Topics == The book begins with a "tutorial" introducing classical algorithm design techniques including backtracking, divide-and-conquer algorithms, and dynamic programming, methods for the analysis of algorithms, and their application in example puzzles. The puzzles themselves are grouped into three sets of 50 puzzles, in increasing order of difficulty. A final two chapters provide brief hints and more detailed solutions to the puzzles, with the solutions forming the majority of pages of the book. Some of the puzzles are well known classics, some are variations of known puzzles making them more algorithmic, and some are new. They include: Puzzles involving chessboards, including the eight queens puzzle, knight's tours, and the mutilated chessboard problem Balance puzzles River crossing puzzles The Tower of Hanoi Finding the missing element in a data stream The geometric median problem for Manhattan distance == Audience and reception == The puzzles in the book cover a wide range of difficulty, and in general do not require more than a high school level of mathematical background. William Gasarch notes that grouping the puzzles only by their difficulty and not by their themes is actually an advantage, as it provides readers with fewer clues about their solutions. Reviewer Narayanan Narayanan recommends the book to any puzzle aficionado, or to anyone who wants to develop their powers of algorithmic thinking. Reviewer Martin Griffiths suggests another group of readers, schoolteachers and university instructors in search of examples to illustrate the power of algorithmic thinking. Gasarch recommends the book to any computer scientist, evaluating it as "a delight".

    Read more →
  • Vector-field consistency

    Vector-field consistency

    Vector-Field Consistency is a consistency model for replicated data (for example, objects), initially described in a paper which was awarded the best-paper prize in the ACM/IFIP/Usenix Middleware Conference 2007. It has since been enhanced for increased scalability and fault-tolerance in a recent paper. == Description == This consistency model was initially designed for replicated data management in ad hoc gaming in order to minimize bandwidth usage without sacrificing playability. Intuitively, it captures the notion that although players require, wish, and take advantage of information regarding the whole of the game world (as opposed to a restricted view to rooms, arenas, etc. of limited size employed in many multiplayer video games), they need to know information with greater freshness, frequency, and accuracy as other game entities are located closer and closer to the player's position. It prescribes a multidimensional divergence bounding scheme, based on a vector field that employs consistency vectors k=(θ,σ,ν), standing for maximum allowed time - or replica staleness, sequence - or missing updates, and value - or user-defined measured replica divergence, applied to all space coordinates in game scenario or world. The consistency vector-fields emanate from field-generators designated as pivots (for example, players) and field intensity attenuates as distance grows from these pivots in concentric or square-like regions. This consistency model unifies locality-awareness techniques employed in message routing and consistency enforcement for multiplayer games, with divergence bounding techniques traditionally employed in replicated database and web scenarios.

    Read more →
  • Documentation science

    Documentation science

    Documentation science is the study of the recording and retrieval of information. It includes methods for storing, retrieving, and sharing of information captured on physical as well as digital documents. This field is closely linked to the fields of library science and information science but has its own theories and practices. The term documentation science was coined by Belgian lawyer and peace activist Paul Otlet. He is considered to be the forefather of information science. He along with Henri La Fontaine laid the foundations of documentation science as a field of study. Professionals in this field are called documentalists. Over the years, documentation science has grown to become a large and important field of study. Evolving from traditional practices like archiving and retrieval to modern theories about the nature of documents, novel methods for organizing digital information, and applications in libraries, research, healthcare, business, and technology and more. This field continues to evolve in the digital age. == Developments in documentation science == 1895: The International Institute of Bibliography (originally Institut International de Bibliographie, IIB) was established on 12 September 1895, in Brussels, Belgium by Paul Otlet and Henri La Fontaine. It aimed to catalog all recorded knowledge using a universal classification system now known as the Universal Decimal Classification (UDC). 1931: International Institute of Bibliography (originally Institut International de Bibliographie, IIB) was renamed The International Institute for Documentation, (Institut International de Documentation, IID). 1934: Paul Otlet envisioned a “radiated library,” a global network of interconnected documents accessible from anywhere via telecommunication. This early idea is now seen as a forerunner of the internet. 1937: American Documentation Institute was founded (1968 nameshift to American Society for Information Science). 1951: Suzanne Briet published Qu'est-ce que la documentation? where she proposed that “a document is evidence in support of a fact,” expanding the definition to include objects such as animals in zoos when they are part of a scientific study. This was a significant theoretical shift in defining documents. 1965-1990: Documentation departments were established, for example, large research libraries, online computer retrieval systems and more. The persons doing the searches were called documentalists. But with the appearance of first CD-ROM databases in the mid-1980s and later the internet in 1990s, these intermediary searches decreased and most such departments closed or merged with other departments. 1996: "Dokvit", Documentation Studies, was established in 1996 at the University of Tromsø in Norway. 2001: The Document Academy was established. It is an international network that celebrates documentation. It was conducted by The Program of Documentation Studies, University of Tromsø, Norway and The School of Information Management and Systems, UC Berkeley. 2003: The first Document Research Conference (DOCAM), a series of conferences made by the Document Academy. DOCAM '03 (2003) was held 13–15 August 2003 at The School of Information Management and Systems (SIMS) at the University of California, Berkeley. 2007: Michael Buckland, Ronald Day, and Birger Hjørland expanded the theoretical foundations of documentation science. They researched and explored documents to be social artifacts, the role of ideology in classification, and how documents influenced knowledge systems. 2010s: The concept of post-documentation or “documentality” began in the 2010s, which focused on how digital traces (e.g., tweets, logs) function as documents without traditional physical form. This led to new thinking in document theory. 2016–present: The Document Academy's DOCAM conferences have continued, offering ongoing developments in the theory and practice of documentation. Themes include affect, memory, activism, and born-digital records. 2017: The journal Information Research published special issues addressing “document theory,” including views on documentation in virtual environments and digital archives. 2020–present: The growth of research data management (RDM) and open science has made documentation practices central to data sharing, metadata standards, and reproducibility in scientific work. == Theoretical foundations == Documentation science has some deep theories that explain what a document is, how people use documents, and how they are organized. These concepts were introduced by scholars who have not only studied libraries, but also philosophy, language, and social sciences. Suzanne Briet described a document as “any material form of evidence” that is made to be used as proof or to share information. An antelope in a zoo, for example, can be a document because it is being studied, classified, and described. Documents are not just things or materials but are also shaped by society. Michael Buckland noted that documents have meaning only when people agree they are useful or valid as information. He explained a document becomes a document when someone decides to use it as evidence. Ronald Day wrote about how documentation is not neutral, it can be influenced by power, ideology, and politics. He claimed that classification systems, like how libraries organize books, are not just technical tools. They also show what kinds of knowledge are seen as more important than others. In recent years, new theories have been introduced, like “documentality” by Maurizio Ferraris. He proposed that a document does not have to be a paper or file, it can also be something digital like a tweet, a database entry, or a log file, as long as it leaves a trace that can be looked at later. This theory helps explain modern digital documents. == Methodologies and practice == Documentation science includes many methods that help people collect, organize, store, and find information. These practices are used in libraries, archives, research labs, companies, and now also in online systems. === Collecting and creating documents === In the past, documentation work included gathering books, articles, reports, and other printed materials. People created records of these materials manually, using catalog cards, indexes, or bibliographies. Paul Otlet’s work with the Universal Bibliographic Repertory is one example. He created millions of card entries to organize knowledge from around the world. Today, documents are not only created by humans. Computers and machines also generate documents, like log files, metadata, and sensor data. These need new tools and methods for collection and management. === Organizing information === Organizing documents has always been a foundational element of documentation science. Methods like classification (dividing things into groups) and indexing (making lists of topics or keywords) help individuals find what they need. A widely used system is the Universal Decimal Classification (UDC) developed by Otlet and La Fontaine. Another is the Library of Congress Classification (LCC) used in the majority of U.S. libraries. Indexing can be performed by humans or by software programs that read the text and add tags to documents. Metadata is also used to describe documents. Metadata is “data about data” like the title, author, date, and subject of a document. Standards like Dublin Core are used in digital libraries to keep metadata consistent. === Retrieval and access === One of the main objectives of documentation is helping users find the right document. This is called information retrieval. In the past, this meant using catalog drawers or printed indexes. Today, people use search engines, databases, and digital libraries. Modern retrieval tools use Boolean logic, ranking algorithms, and sometimes machine learning to show the most useful results first. This is part of what is studied in both documentation science and information retrieval. === Preservation and archiving === Documents require long-term storage. This is called preservation of documents. Printed documents can be damaged by light, pests, or even time on the other hand digital documents can be deemed worthless if formats become outdated or storage facilities fail. Archivists use methods like migration, which includes moving files to new formats, and emulation, which replicates obsolete systems, to preserve materials. These methods and tools are ever changing as new technologies develop. But the main objective of documentation has remained the same, which is to keep information safe, organized, and easy to find. == Documentation in the digital age == With the expansion of the internet, computers, and cloud storage, documents are no longer just books, papers, or reports. They can now be emails, tweets, videos, websites, databases, or even log files created by machines. === Born-digital documents === Many documents today are created directly in digital form. These are called born-digit

    Read more →
  • World model (artificial intelligence)

    World model (artificial intelligence)

    A world model in artificial intelligence is a machine learning system that builds an internal representation of an environment. The model predicts how that environment changes over time in response to actions. Researchers design world models to help agents plan, reason, and act without constant real-world trial and error. World models differ from systems that merely classify or generate outputs. They simulate dynamics such as physics, object interactions, and causality. Early ideas date to the 1990s. Modern versions power robots, autonomous driving, and interactive video generation. == History == Jürgen Schmidhuber introduced the term world model in machine learning in 1990. He proposed recurrent neural networks that predict future states from observations and use those predictions to train agents. David Ha and Schmidhuber revived the concept in a 2018 paper. Their agents learned to drive virtual cars and play video games inside self-generated simulations. Yann LeCun advanced the idea in a 2022 position paper titled "A Path Towards Autonomous Machine Intelligence". He argued that intelligence requires predictive models of the world rather than pure pattern matching. LeCun proposed the joint embedding predictive architecture (JEPA) as a practical foundation. LeCun and collaborators developed several JEPA variants. V-JEPA 2 reached state-of-the-art performance on video understanding and physical reasoning at the time. It supports zero-shot robot control in unfamiliar environments. Introduced in March 2026, LeWorldModel trains stably end-to-end from raw pixels and uses two loss terms and avoids hand-crafted heuristics. LeCun founded Advanced Machine Intelligence Labs in 2026 to further develop world models. Google DeepMind introduced Genie in 2024. The model learned interactive environments from unlabeled internet videos. Genie 2 followed in late 2024 and added three-dimensional generation. The Genie series set benchmarks for general-purpose simulation. Genie 3 was introduced in August 2025. It produces photorealistic, real-time interactive worlds from text prompts which are displayed at 24 frames per second and explored in real time with text or image prompts. The model supports persistent three-dimensional worlds and real-time interaction. Waymo adopted Genie 3 in February 2026 and used it to create a specialized world model for autonomous driving simulation, called the Waymo World Model. It produces synchronized camera and lidar outputs and creates edge cases that real robotaxis rarely encounter. The edge cases were reported to be unusual by PCMag. General Intuition announced a $133.7 million seed round. World Labs raised $1 billion. AMI raised $1.03 billion. In April 2026, Alibaba announced Happy Oyster, its world model designed for real-time and “flowy” world model. It includes a directing mode for world building based on text and image prompts and a wandering mode for exploring the resulting world. It can generate 3-minute in-world video clips. Also in April, World Labs, co-founded by Li Fei Fei, unveiled Spark 2.0, an open-source 3D Gaussian splatting rendering engine that targets smartphone-class devices. In June 2026, Nvidia released Cosmos 3, a family of open-weight models. It combines previously independent physical reasoning, world simulation, and action generation. Cosmos 3 integrates can process and generate text, image, video, audio, and action sequences. The model employs a Mixture-of-Transformers" (MoT) approach. An autoregressive (AR) transformer handles reasoning and next-token prediction, while a diffusion transformer (DT) does multimodal generation. Encoders (ViT for vision, VAE for visual/audio, and domain-specific for actions) and generate a shared representation space using 3D multi-dimensional rotary position embedding (mRoPE) for spatial and temporal information. The family includes Cosmos3-Nano (16B parameters) for workstations; Cosmos3-Super (64B parameters) for research. == Architecture == World models process raw sensory data such as video frames or lidar scans. They compress this input into compact latent representations. The system then predicts future representations rather than pixel-by-pixel reconstructions. Many modern world models use joint embedding predictive architecture (JEPA). An encoder turns observations into embeddings. A predictor estimates one or a suite of embeddings from the current one and an action. In some cases a critic chooses one embedding as the best result. A regularizer keeps embeddings well-behaved. The model trains by minimizing prediction error in embedding space. This approach avoids the high cost of generating every detail. Some architectures add explicit components. A fast reactive path handles immediate responses. A slower deliberative path performs longer-horizon planning. Video prediction accuracy or robot success rates are key metrics, but do not always predict real-world performance. Generative world models such as Genie 3 combine these with a simulator. They accept text prompts or layouts and output consistent video, lidar, or three-dimensional scenes. World models often train with self-supervised learning. They use large unlabeled datasets of video or robot interactions. Self-supervised learning can speed learning. Reinforcement learning can fine-tune a model for specific tasks. == Applications == World models support robot learning. Agents train inside simulations and transfer skills to the physical world. This reduces the need for dangerous or expensive real-world trials. Autonomous vehicles use world models to test rare events. Waymo's system simulates tornadoes or unusual pedestrian behavior. Companies train planners without putting vehicles on public roads. Interactive entertainment benefits from world models. Genie 3 lets users generate playable environments from simple descriptions. Game studios prototype levels faster. Scientific simulation gains from these models. Researchers model physical systems or biological processes at scale. Planners in logistics or urban design test strategies inside accurate digital twins. == Comparison with large language models == Both world models and large language models (LLMs) use inferencing on their inputs to make predictions. LLMs operate on textual inputs. They predict the next token in text sequences. They excel at language-oriented tasks such as translation or summarization. However, they lack understanding of physics. World models operate on sensor inputs such as pixels. They predict state changes in that data in latent space. This design supports planning and causal reasoning. LLMs generate fluent text but often fail at consistent physical predictions. Their architecture employs transformers with refinements such as mixture of experts. World models divide an inferencing task into work performed by encoders, predictors, simulators, and other pieces. They typically handle multimodal inputs such as video, lidar, radar, and audio, guided by textual prompting. LLMs power chatbots and code assistants. World models drive embodied agents that act in dynamic environments, such as autonomous driving. The two may be combined in hybrid systems. For example, a LLM handles instructions, while a world model manages low-level control. World model proponents such as LeCun claim that because LLMs are trained only on text, they have no ability to predict anything beyond text, such as real-world events. == Benchmarks == World model benchmarks test physical understanding, long-term consistency, planning, and generalization from sensor data. Meta introduced three benchmarks for V-JEPA 2. IntPhys 2 measures a model's ability to detect physics violations. It presents pairs of videos that diverge when one breaks physical rules. Humans score near 100% accuracy. V-JEPA 2 achieves little better than random chance on many conditions. Minimal Video Pairs (MVPBench) tests physical understanding through multiple-choice questions based on short video clips. It probes object interactions and causality. Something-Something tests action recognition. Epic-Kitchens-100 tests human action anticipation. DeepMind benchmark: Interactive evaluation measures consistency over minutes of interaction, memory of off-screen objects, and response to user actions or text prompts. Waymo benchmark: Output generation quality: Metrics include realism, controllability (via text prompts), and usefulness for training planners in simulated worlds. However, pixel reconstruction error rate with episodic rewards often fails. Other: Epic-Kitchens-100 (often measured with Recall@5) Ego4D 50 Salads, Breakfast, etc. Potential benchmarks: Zero-shot transfer to robots Long-horizon planning Implausible prediction rate

    Read more →
  • Literature review

    Literature review

    A literature review is an overview of previously published works on a particular topic. The term can refer to a full scholarly paper or a section of a scholarly work such as books or articles. Either way, a literature review provides the researcher/author and the audiences with general information of an existing knowledge of a particular topic. A good literature review has a proper research question, a proper theoretical framework, and/or a chosen research method. It serves to situate the current study within the body of the relevant literature and provides context for the reader. In such cases, the review usually precedes the methodology and results sections of the work. Producing a literature review is often part of a graduate and post-graduate requirement, included in the preparation of a thesis, dissertation, or a journal article. Literature reviews are also common in a research proposal or prospectus (the document approved before a student formally begins a dissertation or thesis). A literature review can be a type of a review article. In this sense, it is a scholarly paper that presents the current knowledge including substantive findings as well as theoretical and methodological contributions to a particular topic. Literature reviews are secondary sources and do not report new or original experimental work. Most often associated with academic-oriented literature, such reviews are found in academic journals and are not to be confused with book reviews, which may also appear in the same publication. Literature reviews are a basis for research in nearly every academic field. == Types == Since the concept of a systematic review was formalized in the 1970s, a basic division among types of reviews is the dichotomy of narrative reviews versus systematic reviews. The main types of narrative reviews are evaluative, exploratory, and instrumental. A fourth type of review of literature (the scientific literature) is the systematic review but it is not called a literature review, which absent further specification, conventionally refers to narrative reviews. A systematic review focuses on a specific research question to identify, appraise, select, and synthesize all high-quality research evidence and arguments relevant to that question. A meta-analysis is typically a systematic review using statistical methods to effectively combine the data used on all selected studies to produce a more reliable result. Torraco (2016) describes an integrative literature review. The purpose of an integrative literature review is to generate new knowledge on a topic through the process of review, critique, and synthesis of the literature under investigation. George et al (2023) offer an extensive overview of review approaches. They also propose a model for selecting an approach by looking at the purpose, object, subject, community, and practices of the review. They describe six different types of review, each with their own unique purposes: Exploratory or scoping reviews focus on breadth as opposed to depth Systematic or integrative reviews integrate empirical studies on a topic Meta-narrative reviews are qualitative and use literature to compare research or practice communities Problematizing or critical reviews propose new perspectives on a concept by association with other literature Meta-analyses and meta-regressions integrate quantitative studies and identify moderators Mixed research syntheses combine other review approaches in the same paper == Process and product == Shields and Rangarajan (2013) distinguish between the process of reviewing the literature and a finished work or product known as a literature review. The process of reviewing the literature is often ongoing and informs many aspects of the empirical research project. The process of reviewing the literature requires different kinds of activities and ways of thinking. Shields and Rangarajan (2013) and Granello (2001) link the activities of doing a literature review with Benjamin Bloom's revised taxonomy of the cognitive domain (ways of thinking: remembering, understanding, applying, analyzing, evaluating, and creating). === Use of artificial intelligence in a literature review === Artificial intelligence (AI) is reshaping traditional literature reviews across various disciplines. Generative pre-trained transformers, such as ChatGPT, are often used by students and academics for review purposes. Since 2023, an increasing number of tools powered by large language models and other artificial intelligence technologies have been developed to assist, automate, or generate literature reviews. Nevertheless, the employment of ChatGPT in academic reviews is problematic due to ChatGPT's propensity to "hallucinate". In response, efforts are being made to mitigate these hallucinations through the integration of plugins. For instance, Rad et al. (2023) used ScholarAI for review in cardiothoracic surgery.

    Read more →
  • Golden record (informatics)

    Golden record (informatics)

    In informatics, a golden record is the valid version of a data element (record) in a single source of truth system. It may refer to a database, specific table or data field, or any unit of information used. A golden copy is a consolidated data set, and is supposed to provide a single source of truth and a "well-defined version of all the data entities in an organizational ecosystem". Other names sometimes used include master source or master version. The term has been used in conjunction with data quality, master data management, and similar topics. (Different technical solutions exist, see master data management). == Master data == In master data management (MDM), the golden copy refers to the master data (master version) of the reference data which works as an authoritative source for the "truth" for all applications in a given IT landscape.

    Read more →
  • Artificial intelligence in Brazilian industry

    Artificial intelligence in Brazilian industry

    In 2022, 16.9% (1,620) of the 9,586 Brazilian industrial companies with 100 or more employees used artificial intelligence in their operations Among the companies that used AI, the areas of administration (73.8%), product project development (65.9%), processes, services and marketing (65.1%) were those that used it the most, followed by the areas of production (56.4%) and logistics (48.4%). == Current scenario == === Adoption in Brazilian industrial sectors === In senior management, the majority (56%) of executives have a long-term vision for its use. The study also shows that IT, Innovation, and Marketing are the areas where AI use is most widespread, and that 43% of companies are developing or adapting the algorithms they use. The majority of large institutions that reported some type of AI use purchased these solutions from other companies (76%). Some factors for the adoption of artificial intelligence in companies include the establishment of an autonomous strategy by the company (87.0%), and the influence of suppliers and/or customers (63.0%) and the main difficulties in using technologies were high costs (80.8%), lack of qualified personnel in the company (54.6%) and excessive economic risks (49.5%). Three variables are considered the most relevant to explain the option to use AI: the implementation of a digital security policy, the size of companies with 250 or more employees and the characteristics of the company related to information and communication. When analyzing AI use by company size in Brazil, large companies have the highest proportion of AI use, mainly due to their investment capacity and technology experimentation. However, when comparing Brazil and Europe, indicators show an acceleration in AI use among large European companies, while in Brazil the situation remains stable. In 2023, 30% of large companies in the European bloc used some type of AI, a figure that rose to 41% in 2024, while in Brazil these proportions were 41% in 2023 and 38% in 2024. === Workforce === The challenge of upskilling begins with employees who are capable of understanding recent technological changes. Similarly, companies must create the environment and conditions for workforce development conducive to innovation, and universities must be prepared to provide knowledge aligned with the transition process, which in turn must be supported by public policies. The concern with training a specialized workforce in AI can be seen in the low number of graduates and PhDs in computer science and computer engineering in Brazil, compared to the number shown in other countries. As recorded in the document Recommendations for the Advancement of Artificial Intelligence in Brazil, 2019 data from the Coordination for the Improvement of Higher Education Personnel (CAPES) indicate that "the number of PhDs graduated annually in computing remained below 400 in 2016, and is not expected to have increased during the Covid-19 pandemic" (ABC, 2023). In the United States, by contrast, the number of PhDs graduated in these two areas has remained around 1,800 for the past 11 years, and during this period, the number of PhDs specializing in AI jumped from 10% to 19%. Based on data from the CNPq Lattes Platform (October 2019), it is possible to observe that the number of professionals in the AI field in Brazil is 4,429 specialists. This is still a small number compared to the 415,166 IT jobs in the country's business sector alone. === R&D, scientific production and integration with industry === China and the United States lead in the number of publications. These two countries are followed by the G7 members: India, Austria, South Korea, and Spain. Brazil appears in the next group, alongside the Netherlands, Russia, Indonesia, and Ireland. Regarding the promotion of research and technologies related to AI, public entities such as the Coordination for the Improvement of Higher Education Personnel (Capes) and the National Council for Scientific and Technological Development (CNPq) stood out as the main funders. Currently, different countries and territories have been promoting the development of Artificial Intelligence (AI). In the Brazilian case, one of the main initiatives is the creation of Engineering Research Centers/Applied Research Centers (CPE/CPA) in AI by the São Paulo Research Foundation (FAPESP), in collaboration with the Ministry of Science, Technology and Innovation (MCTI), the Ministry of Communications (MC) and the Brazilian Internet Steering Committee (CGI.br). In terms of the number of patents filed and the volume of investments, the leading nations in AI are the United States, China, France, Germany, the United Kingdom, Russia, India, Switzerland, Japan, South Korea, the Netherlands, Sweden, Finland, Ireland, Singapore, Canada, Israel, and Italy. Brazil appears among the top twenty countries in some rankings, mainly due to its good number of publications (approximately 10% of the number of articles published by the United States). The US is home to approximately 60% of the world's top AI researchers, followed by China (11%), Europe (10%), and Canada (6%). To change this scenario, in August 2024, the Brazilian government announced an investment of R$23 billion until 2028 in artificial intelligence, seeking to “transform the country into a global reference in innovation”. == Future challenges == The Organization for Economic Cooperation and Development (2020) report highlighted three factors that hinder the digital transformation journey and application of AI in Brazil: insufficient infrastructure, high costs due to the tax system, and financial limitations, such as limited access to financing. The costs of adopting technology, its incompatibility with the business, and the lack of training also represent obstacles that Brazilian industry must overcome. There are also inherent obstacles for companies. A McKinsey review emphasizes that once a company chooses one or more sectors to focus on, it must select specific applications. Buyers aren't interested in artificial intelligence simply because it's a breakthrough technology; they want AI to generate a good return on investment, whether by solving specific problems, saving money, or increasing sales. If an AI vendor tried to offer a horizontal solution, the value proposition might not be as compelling. Part of the solution to Brazil's technological backwardness involves building an ecosystem fueled by private institutions, universities, and governments.

    Read more →
  • BERT (language model)

    BERT (language model)

    Bidirectional encoder representations from transformers (BERT) is a language model introduced in October 2018 by researchers at Google. It learns to represent text as a sequence of vectors using self-supervised learning. It uses the encoder-only transformer architecture. BERT dramatically improved the state of the art for large language models. As of 2020, BERT is a ubiquitous baseline in natural language processing (NLP) experiments. BERT is trained by masked token prediction and next sentence prediction. With this training, BERT learns contextual, latent representations of tokens in their context, similar to ELMo and GPT-2. It found applications for many natural language processing tasks, such as coreference resolution and polysemy resolution. It improved on ELMo and spawned the study of "BERTology", which attempts to interpret what is learned by BERT. BERT was originally implemented in the English language at two model sizes, BERTBASE (110 million parameters) and BERTLARGE (340 million parameters). Both were trained on the Toronto BookCorpus (800M words) and English Wikipedia (2,500M words). The weights were released on GitHub. On March 11, 2020, 24 smaller models were released, the smallest being BERTTINY with just 4 million parameters. == Architecture == BERT is an "encoder-only" transformer architecture. At a high level, BERT consists of 4 modules: Tokenizer: This module converts a piece of English text into a sequence of integers ("tokens"). Embedding: This module converts the sequence of tokens into an array of real-valued vectors representing the tokens. It represents the conversion of discrete token types into a lower-dimensional Euclidean space. Encoder: a stack of Transformer blocks with self-attention, but without causal masking. Task head: This module converts the final representation vectors into one-shot encoded tokens again by producing a predicted probability distribution over the token types. It can be viewed as a simple decoder, decoding the latent representation into token types, or as an "un-embedding layer". The task head is necessary for pre-training, but it is often unnecessary for so-called "downstream tasks," such as question answering or sentiment classification. Instead, one removes the task head and replaces it with a newly initialized module suited for the task, and finetune the new module. The latent vector representation of the model is directly fed into this new module, allowing for sample-efficient transfer learning. === Embedding === This section describes the embedding used by BERTBASE. The other one, BERTLARGE, is similar, just larger. The tokenizer of BERT is WordPiece, which is a sub-word strategy like byte-pair encoding. Its vocabulary size is 30,000, and any token not appearing in its vocabulary is replaced by [UNK] ("unknown"). The first layer is the embedding layer, which contains three components: token type embeddings, position embeddings, and segment type embeddings. Token type: The token type is a standard embedding layer, translating a one-hot vector into a dense vector based on its token type. Position: The position embeddings are based on a token's position in the sequence. BERT uses absolute position embeddings, where each position in a sequence is mapped to a real-valued vector. Each dimension of the vector consists of a sinusoidal function that takes the position in the sequence as input. Segment type: Using a vocabulary of just 0 or 1, this embedding layer produces a dense vector based on whether the token belongs to the first or second text segment in that input. In other words, type-1 tokens are all tokens that appear after the [SEP] special token. All prior tokens are type-0. The three embedding vectors are added together representing the initial token representation as a function of these three pieces of information. After embedding, the vector representation is normalized using a LayerNorm operation, outputting a 768-dimensional vector for each input token. After this, the representation vectors are passed forward through 12 Transformer encoder blocks, and are decoded back to 30,000-dimensional vocabulary space using a basic affine transformation layer. === Architectural family === The encoder stack of BERT has 2 free parameters: L {\displaystyle L} , the number of layers, and H {\displaystyle H} , the hidden size. There are always H / 64 {\displaystyle H/64} self-attention heads, and the feed-forward/filter size is always 4 H {\displaystyle 4H} . By varying these two numbers, one obtains an entire family of BERT models. For BERT: the feed-forward size and filter size are synonymous. Both of them denote the number of dimensions in the middle layer of the feed-forward network. the hidden size and embedding size are synonymous. Both of them denote the number of real numbers used to represent a token. The notation for encoder stack is written as L/H. For example, BERTBASE is written as 12L/768H, BERTLARGE as 24L/1024H, and BERTTINY as 2L/128H. == Training == === Pre-training === BERT was pre-trained simultaneously on two tasks: Masked language modeling (MLM): In this task, BERT ingests a sequence of words, where one word may be randomly changed ("masked"), and BERT tries to predict the original words that had been changed. For example, in the sentence "The cat sat on the [MASK]," BERT would need to predict "mat." This helps BERT learn bidirectional context, meaning it understands the relationships between words not just from left to right or right to left but from both directions at the same time. Next sentence prediction (NSP): In this task, BERT is trained to predict whether one sentence logically follows another. For example, given two sentences, "The cat sat on the mat" and "It was a sunny day", BERT has to decide if the second sentence is a valid continuation of the first one. This helps BERT understand relationships between sentences, which is important for tasks like question answering or document classification. ==== Masked language modeling ==== In masked language modeling, 15% of tokens would be randomly selected for masked-prediction task, and the training objective was to predict the masked token given its context. In more detail, the selected token is: replaced with a [MASK] token with probability 80%, replaced with a random word token with probability 10%, not replaced with probability 10%. The reason not all selected tokens are masked is to avoid the dataset shift problem. The dataset shift problem arises when the distribution of inputs seen during training differs significantly from the distribution encountered during inference. A trained BERT model might be applied to word representation (like Word2Vec), where it would be run over sentences not containing any [MASK] tokens. It is later found that more diverse training objectives are generally better. As an illustrative example, consider the sentence "my dog is cute". It would first be divided into tokens like "my1 dog2 is3 cute4". Then a random token in the sentence would be picked. Let it be the 4th one "cute4". Next, there would be three possibilities: with probability 80%, the chosen token is masked, resulting in "my1 dog2 is3 [MASK]4"; with probability 10%, the chosen token is replaced by a uniformly sampled random token, such as "happy", resulting in "my1 dog2 is3 happy4"; with probability 10%, nothing is done, resulting in "my1 dog2 is3 cute4". After processing the input text, the model's 4th output vector is passed to its decoder layer, which outputs a probability distribution over its 30,000-dimensional vocabulary space. ==== Next sentence prediction ==== Given two sentences, the model predicts if they appear sequentially in the training corpus, outputting either [IsNext] or [NotNext]. During training, the algorithm sometimes samples two sentences from a single continuous span in the training corpus, while at other times, it samples two sentences from two discontinuous spans. The first sentence starts with a special token, [CLS] (for "classify"). The two sentences are separated by another special token, [SEP] (for "separate"). After processing the two sentences, the final vector for the [CLS] token is passed to a linear layer for binary classification into [IsNext] and [NotNext]. For example: Given "[CLS] my dog is cute [SEP] he likes playing [SEP]", the model should predict [IsNext]. Given "[CLS] my dog is cute [SEP] how do magnets work [SEP]", the model should predict [NotNext]. === Fine-tuning === BERT is meant as a general pretrained model for various applications in natural language processing. That is, after pre-training, BERT can be fine-tuned with fewer resources on smaller datasets to optimize its performance on specific tasks such as natural language inference and text classification, and sequence-to-sequence-based language generation tasks such as question answering and conversational response generation. The original BERT paper published results demonstrating that a small amount of fine

    Read more →
  • Single source of truth

    Single source of truth

    In information science and information technology, single source of truth (SSOT) architecture, or single point of truth (SPOT) architecture, for information systems is the practice of structuring information models and associated data schemas such that every data element is mastered (or edited) in only one place, providing data normalization to a canonical form (for example, in database normalization or content transclusion). There are several scenarios with respect to copies and updates: The master data is never copied and instead only references to it are made; this means that all reads and updates go directly to the SSOT. The master data is copied but the copies are only read and only the master data is updated; if requests to read data are only made on copies, this is an instance of CQRS. The master data is copied and the copies are updated; this needs a reconciliation mechanism when there are concurrent updates. Updates on copies can be thrown out whenever a concurrent update is made on the master, so they are not considered fully committed until propagated to the master. (many blockchains work that way.) Concurrent updates are merged. (if an automatic merge fails, it could fall back on another strategy, which could be the previous strategy or something else like manual intervention, which most source version control systems do.) The advantages of SSOT architectures include easier prevention of mistaken inconsistencies (such as a duplicate value/copy somewhere being forgotten), and greatly simplified version control. Without a SSOT, dealing with inconsistencies implies either complex and error-prone consensus algorithms, or using a simpler architecture that's liable to lose data in the face of inconsistency (the latter may seem unacceptable but it is sometimes a very good choice; it is how most blockchains operate: a transaction is actually final only if it was included in the next block that is mined). Ideally, SSOT systems provide data that are authentic (and authenticatable), relevant, and referable. Deployment of an SSOT architecture is becoming increasingly important in enterprise settings where incorrectly linked duplicate or de-normalized data elements (a direct consequence of intentional or unintentional denormalization of any explicit data model) pose a risk for retrieval of outdated, and therefore incorrect, information. Common examples (i.e., example classes of implementation) are as follows: In electronic health records (EHRs), it is imperative to accurately validate patient identity against a single referential repository, which serves as the SSOT. Duplicate representations of data within the enterprise would be implemented by the use of pointers rather than duplicate database tables, rows, or cells. This ensures that data updates to elements in the authoritative location are comprehensively distributed to all federated database constituencies in the larger overall enterprise architecture. EHRs are an excellent class for exemplifying how SSOT architecture is both poignantly necessary and challenging to achieve: it is challenging because inter-organization health information exchange is inherently a cybersecurity competence hurdle, and nonetheless it is necessary, to prevent medical errors, to prevent the wasted costs of inefficiency (such as duplicated work or rework), and to make the primary care and medical home concepts feasible (to achieve competent care transitions). Single-source publishing as a general principle or ideal in content management relies on having SSOTs, via transclusion or (otherwise, at least) substitution. Substitution happens via libraries of objects that can be propagated as static copies which are later refreshed when necessary (that is, when refreshing of the copy-paste or import is triggered by a larger updating event). Component content management systems are a class of content management systems that aim to provide competence on this level. == Implementation == === Ontologic interactions === An acknowledged prerequisite (of the notion that any given single source of truth can exist) is that it depends on the ontologic condition that no more than a single truth (about any particular fact or idea) exists, an assertion that is ontologic in both the IT sense and the general sense of that word. In many instances, this presents no problem (for example, within particular namespaces, or even across them, as long as naming collisions or broader name conflicts are adequately handled). The broadest contexts (and thus thorniest, regarding ontologic discrepancies) require adequate epistemic regime comparison and reconciliation (or at least negotiation or transactional exchanges). An archetypal example of this class of reconciliation is that two theological seminary libraries, from two different religions (X and Y), could exchange information with an SSOT architecture, but the unification of truth would reside on the level of the statement that "religion X asserts that God is purple whereas religion Y asserts that God is green", rather than on the level of "God is purple" or "God is green". === Architectures or architectural features === An ideal implementation of SSOT is rarely possible in most enterprises. This is because many organisations have multiple information systems, each of which needs access to data relating to the same entities (e.g., customer). Often these systems are purchased as commercial off-the-shelf products from vendors and cannot be modified in trivial ways. Each of these various systems therefore needs to store its own version of common data or entities, and therefore each system must retain its own copy of a record (hence immediately violating the SSOT approach defined above). For example, an enterprise resource planning (ERP) system (such as SAP or Oracle e-Business Suite) may store a customer record; the customer relationship management (CRM) system also needs a copy of the customer record (or part of it) and the warehouse dispatch system might also need a copy of some or all of the customer data (e.g., shipping address). In cases where vendors do not support such modifications, it is not always possible to replace these records with pointers to the SSOT. For organisations (with more than one information system) wishing to implement a Single Source of Truth (without modifying all but one master system to store pointers to other systems for all entities), some supporting architectures are: Master data management (MDM) Event store and event sourcing (ES) ==== Master data management (MDM) ==== A master data management system typically serves as the source of truth for an organization's metadata, helping to ensure accuracy and consistency throughout that organizations multiple data sources. Typically the MDM acts as a hub for multiple systems, many of which could allow (be the source of truth for) updates to different aspects of information on a given entity. For example, the CRM system may be the "source of truth" for most aspects of the customer, and is updated by a call centre operator. However, a customer may (for example) also update their address via a customer service web site, with a different back-end database from the CRM system. The MDM application receives updates from multiple sources, acts as a broker to determine which updates are to be regarded as authoritative (the golden record) and then syndicates this updated data to all subscribing systems. The MDM application normally requires an ESB to syndicate its data to multiple subscribing systems. ==== Event store and event sourcing (ES) ==== In event oriented architectures, it has become increasingly common to find an implementation of the Event Sourcing pattern which stores the system state as an ordered sequence of state changes. To do this, you need an Event Store, a particular type of database designed to hold all the events that change the state of the system. The event store in an Event Sourcing + Command Query Responsibility Separation + Domain Driven Design + Messaging architecture is in fact a "single source of truth", with the additional advantage that it can also act as an Enterprise Service Bus as it can listen directly to the event store for status changes as everything passes by. In addition, by saving all the events, it also plays the role of Data Warehouse. One last advantage is that through this system the Shared Database pattern can be implemented, another technique not mentioned to obtain a single source of truth. ==== Data warehouse (DW) ==== While the primary purpose of a data warehouse is to support reporting and analysis of data that has been combined from multiple sources, the fact that such data has been combined (according to business logic embedded in the data transformation and integration processes) means that the data warehouse is often used as a de facto SSOT. Generally, however, the data available from the data warehouse are not used to update other systems; rather the DW becomes

    Read more →
  • Information seeking

    Information seeking

    Information seeking is the process or activity of attempting to obtain information in both human and technological contexts. Information seeking is related to, but different from, information retrieval (IR). == Compared to information retrieval == Traditionally, IR tools have been designed for IR professionals to enable them to effectively and efficiently retrieve information from a source. It is assumed that the information exists in the source and that a well-formed query will retrieve it (and nothing else). It has been argued that laypersons' information seeking on the internet is very different from information retrieval as performed within the IR discourse. Yet, internet search engines are built on IR principles. Since the late 1990s a body of research on how casual users interact with internet search engines has been forming, but the topic is far from fully understood. IR can be said to be technology-oriented, focusing on algorithms and issues such as precision and recall. Information seeking may be understood as a more human-oriented and open-ended process than information retrieval. In information seeking, one does not know whether there exists an answer to one's query, so the process of seeking may provide the learning required to satisfy one's information need. == In different contexts == Much library and information science (LIS) research has focused on the information-seeking practices of practitioners within various fields of professional work. Studies have been carried out into the information-seeking behaviors of librarians, academics, medical professionals, engineers, lawyers and mini-publics(among others). Much of this research has drawn on the work done by Leckie, Pettigrew (now Fisher) and Sylvain, who in 1996 conducted an extensive review of the LIS literature (as well as the literature of other academic fields) on professionals' information seeking. The authors proposed an analytic model of professionals' information seeking behaviour, intended to be generalizable across the professions, thus providing a platform for future research in the area. The model was intended to "prompt new insights... and give rise to more refined and applicable theories of information seeking" (1996, p. 188). The model has been adapted by Wilkinson (2001) who proposes a model of the information seeking of lawyers. Recent studies in this topic address the concept of information-gathering that "provides a broader perspective that adheres better to professionals' work-related reality and desired skills." (Solomon & Bronstein, 2021). == Theories of information-seeking behavior == A variety of theories of information behavior – e.g. Zipf's Principle of Least Effort, Brenda Dervin's Sense Making, Elfreda Chatman's Life in the Round – seek to understand the processes that surround information seeking. In addition, many theories from other disciplines have been applied in investigating an aspect or whole process of information seeking behavior. A review of the literature on information seeking behavior shows that information seeking has generally been accepted as dynamic and non-linear (Foster, 2005; Kuhlthau 2006). People experience the information search process as an interplay of thoughts, feelings and actions (Kuhlthau, 2006). Donald O. Case (2007) also wrote a good book that is a review of the literature. Information seeking has been found to be linked to a variety of interpersonal communication behaviors beyond question-asking, to include strategies such as candidate answers. Robinson's (2010) research suggests that when seeking information at work, people rely on both other people and information repositories (e.g., documents and databases), and spend similar amounts of time consulting each (7.8% and 6.4% of work time, respectively; 14.2% in total). However, the distribution of time among the constituent information seeking stages differs depending on the source. When consulting other people, people spend less time locating the information source and information within that source, similar time understanding the information, and more time problem solving and decision making, than when consulting information repositories. Furthermore, the research found that people spend substantially more time receiving information passively (i.e., information that they have not requested) than actively (i.e., information that they have requested), and this pattern is also reflected when they provide others with information. == Wilson's nested model of conceptual areas == The concepts of information seeking, information retrieval, and information behaviour are objects of investigation of information science. Within this scientific discipline a variety of studies has been undertaken analyzing the interaction of an individual with information sources in case of a specific information need, task, and context. The research models developed in these studies vary in their level of scope. Wilson (1999) therefore developed a nested model of conceptual areas, which visualizes the interrelation of the here mentioned central concepts. Wilson defines models of information behavior to be "statements, often in the form of diagrams, that attempt to describe an information-seeking activity, the causes and consequences of that activity, or the relationships among stages in information-seeking behaviour" (1999: 250).

    Read more →