AI Generator Website

AI Generator Website — independent reviews, comparisons, pricing and step-by-step guides on Aizhi.

ImageNet

The ImageNet project is a large visual database designed for use in visual object recognition software research. More than 14 million images have been hand-annotated by the project to indicate what objects are pictured and in at least one million of the images, bounding boxes are also provided. ImageNet contains more than 20,000 categories, with a typical category, such as "balloon" or "strawberry", consisting of several hundred images. The database of annotations of third-party image URLs is freely available directly from ImageNet, though the actual images are not owned by ImageNet. Since 2010, the ImageNet project runs an annual software contest, the ImageNet Large Scale Visual Recognition Challenge (ILSVRC), where software programs compete to correctly classify and detect objects and scenes. The challenge uses a "trimmed" list of one thousand non-overlapping classes. == History == AI researcher Fei-Fei Li began working on the idea for ImageNet in 2006. At a time when most AI research focused on models and algorithms, Li wanted to expand and improve the data available to train AI algorithms. In 2007, Li met with Princeton professor Christiane Fellbaum, one of the creators of WordNet, to discuss the project. As a result of this meeting, Li went on to build ImageNet starting from the roughly 22,000 nouns of WordNet and using many of its features. She was also inspired by a 1987 estimate that the average person recognizes roughly 30,000 different kinds of objects. As an assistant professor at Princeton, Li assembled a team of researchers to work on the ImageNet project. They used Amazon Mechanical Turk to help with the classification of images. Labeling started in July 2008 and ended in April 2010. It took 49K workers from 167 countries filtering and labeling over 160M candidate images. They had enough budget to have each of the 14 million images labelled three times. The original plan called for 10,000 images per category, for 40,000 categories at 400 million images, each verified 3 times. They found that humans can classify at most 2 images/sec. At this rate, it was estimated to take 19 human-years of labor (without rest). They presented their database for the first time as a poster at the 2009 Conference on Computer Vision and Pattern Recognition (CVPR) in Florida, titled "ImageNet: A Preview of a Large-scale Hierarchical Dataset". The poster was reused at Vision Sciences Society 2009. In 2009, Alex Berg suggested adding object localization as a task. Li approached PASCAL Visual Object Classes contest in 2009 for a collaboration. It resulted in the subsequent ImageNet Large Scale Visual Recognition Challenge starting in 2010, which has 1000 classes and object localization, as compared to PASCAL VOC which had just 20 classes and 19,737 images (in 2010). === Significance for deep learning === On 30 September 2012, a convolutional neural network (CNN) called AlexNet achieved a top-5 error of 15.3% in the ImageNet 2012 Challenge, more than 10.8 percentage points lower than that of the runner-up. Using convolutional neural networks was feasible due to the use of graphics processing units (GPUs) during training, an essential ingredient of the deep learning revolution. According to The Economist, "Suddenly people started to pay attention, not just within the AI community but across the technology industry as a whole." In 2015, AlexNet was outperformed by Microsoft's very deep CNN with over 100 layers, which won the ImageNet 2015 contest, having 3.57% error on the test set. Andrej Karpathy estimated in 2014 that with concentrated effort, he could reach 5.1% error rate, and ~10 people from his lab reached ~12-13% with less effort. It was estimated that with maximal effort, a human could reach 2.4%. == Dataset == ImageNet crowdsources its annotation process. Image-level annotations indicate the presence or absence of an object class in an image, such as "there are tigers in this image" or "there are no tigers in this image". Object-level annotations provide a bounding box around the (visible part of the) indicated object. ImageNet uses a variant of the broad WordNet schema to categorize objects, augmented with 120 categories of dog breeds to showcase fine-grained classification. In 2012, ImageNet was the world's largest academic user of Mechanical Turk. The average worker identified 50 images per minute. The original plan of the full ImageNet would have roughly 50M clean, diverse and full resolution images spread over approximately 50K synsets. This was not achieved. The summary statistics given on April 30, 2010: Total number of non-empty synsets: 21841 Total number of images: 14,197,122 Number of images with bounding box annotations: 1,034,908 Number of synsets with SIFT features: 1000 Number of images with SIFT features: 1.2 million === Categories === The categories of ImageNet were filtered from the WordNet concepts. Each concept, since it can contain multiple synonyms (for example, "kitty" and "young cat"), so each concept is called a "synonym set" or "synset". There were more than 100,000 synsets in WordNet 3.0, majority of them are nouns (80,000+). The ImageNet dataset filtered these to 21,841 synsets that are countable nouns that can be visually illustrated. Each synset in WordNet 3.0 has a "WordNet ID" (wnid), which is a concatenation of part of speech and an "offset" (a unique identifying number). Every wnid starts with "n" because ImageNet only includes nouns. For example, the wnid of synset "dog, domestic dog, Canis familiaris" is "n02084071". The categories in ImageNet fall into 9 levels, from level 1 (such as "mammal") to level 9 (such as "German shepherd"). === Image format === The images were scraped from online image search (Google, Picsearch, MSN, Yahoo, Flickr, etc) using synonyms in multiple languages. For example: German shepherd, German police dog, German shepherd dog, Alsatian, ovejero alemán, pastore tedesco, 德国牧羊犬. ImageNet consists of images in RGB format with varying resolutions. For example, in ImageNet 2012, "fish" category, the resolution ranges from 4288 x 2848 to 75 x 56. In machine learning, these are typically preprocessed into a standard constant resolution, and whitened, before further processing by neural networks. For example, in PyTorch, ImageNet images are by default normalized by dividing the pixel values so that they fall between 0 and 1, then subtracting by [0.485, 0.456, 0.406], then dividing by [0.229, 0.224, 0.225]. These are the mean and standard deviations for ImageNet, so this whitens the input data. === Labels and annotations === Each image is labelled with exactly one wnid. Dense SIFT features (raw SIFT descriptors, quantized codewords, and coordinates of each descriptor/codeword) for ImageNet-1K were available for download, designed for bag of visual words. The bounding boxes of objects were available for about 3000 popular synsets with on average 150 images in each synset. Furthermore, some images have attributes. They released 25 attributes for ~400 popular synsets: Color: black, blue, brown, gray, green, orange, pink, red, violet, white, yellow Pattern: spotted, striped Shape: long, round, rectangular, square Texture: furry, smooth, rough, shiny, metallic, vegetation, wooden, wet === ImageNet-21K === The full original dataset is referred to as ImageNet-21K. ImageNet-21k contains 14,197,122 images divided into 21,841 classes. Some papers round this up and name it ImageNet-22k. The full ImageNet-21k was released in Fall of 2011, as fall11_whole.tar. There is no official train-validation-test split for ImageNet-21k. Some classes contain only 1-10 samples, while others contain thousands. === ImageNet-1K === There are various subsets of the ImageNet dataset used in various context, sometimes referred to as "versions". One of the most highly used subsets of ImageNet is the "ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2012–2017 image classification and localization dataset". This is also referred to in the research literature as ImageNet-1K or ILSVRC2017, reflecting the original ILSVRC challenge that involved 1,000 classes. ImageNet-1K contains 1,281,167 training images, 50,000 validation images and 100,000 test images. Each category in ImageNet-1K is a leaf category, meaning that there are no child nodes below it, unlike ImageNet-21K. For example, in ImageNet-21K, there are some images categorized as simply "mammal", whereas in ImageNet-1K, there are only images categorized as things like "German shepherd", since there are no child-words below "German shepherd". === Later developments === In the WordNet they built ImageNet on, there were 2832 synsets in the "person" subtree. During 2018--2020 period, they removed the download of the ImageNet-21k as they went through extensive filtering in these person synsets. Out of these 2832 synsets, 1593 were deemed "potentially offensive". Out of the remaining 1239, 1081 were deemed not really "visual". The result was that only 158 syn
Read more →
FAIR data

FAIR data is data which meets the 2016 FAIR principles of findability, accessibility, interoperability, and reusability (FAIR). The FAIR principles emphasize machine-actionability (i.e., the capacity of computational systems to find, access, interoperate, and reuse data with none or minimal human intervention) because humans increasingly rely on computational support to deal with data as a result of the increase in the volume, complexity, and rate of production of data. The abbreviation FAIR/O data is sometimes used to indicate that the dataset or database in question complies with the FAIR principles and also carries an explicit data‑capable open license. == FAIR principles published by GO FAIR == Findable The first step in (re)using data is to find them. Metadata and data should be easy to find for both humans and computers. Machine-readable metadata are essential for automatic discovery of datasets and services, so this is an essential component of the FAIRification process. F1. (Meta)data are assigned a globally unique and persistent identifier F2. Data are described with rich metadata (defined by R1 below) F3. Metadata clearly and explicitly include the identifier of the data they describe F4. (Meta)data are registered or indexed in a searchable resource Accessible Once the user finds the required data, they need to know how they can be accessed, possibly including authentication and authorisation. A1. (Meta)data are retrievable by their identifier using a standardised communications protocol A1.1 The protocol is open, free, and universally implementable A1.2 The protocol allows for an authentication and authorisation procedure, where necessary A2. Metadata are accessible, even when the data are no longer available Interoperable The data usually need to be integrated with other data. In addition, the data need to interoperate with applications or workflows for analysis, storage, and processing. I1. (Meta)data use a formal, accessible, shared, and broadly applicable language for knowledge representation I2. (Meta)data use vocabularies that follow FAIR principles I3. (Meta)data include qualified references to other (meta)data Reusable The ultimate goal of FAIR is to optimise the reuse of data. To achieve this, metadata and data should be well-described so that they can be replicated and/or combined in different settings. R1. (Meta)data are richly described with a plurality of accurate and relevant attributes R1.1. (Meta)data are released with a clear and accessible data usage license R1.2. (Meta)data are associated with detailed provenance R1.3. (Meta)data meet domain-relevant community standards The principles refer to three types of entities: data (or any digital object), metadata (information about that digital object), and infrastructure. For instance, principle F4 defines that both metadata and data are registered or indexed in a searchable resource (the infrastructure component). === Acceptance and implementation === Before FAIR, a 2007 OECD report was the most influential paper discussing similar ideas related to data accessibility. In January 2014, the Lorentz Centre at Leiden University hosted a workshop entitled "Jointly designing a data FAIRPORT" where the participants first formulated the FAIR principles. After further discussions, they were published in the March 2016 issue of Scientific Data. At the 2016 G20 Hangzhou summit, the G20 leaders issued a statement endorsing the application of FAIR principles to research. Also in 2016, a group of Australian organisations developed a Statement on FAIR Access to Australia's Research Outputs, which aimed to extend the principles to research outputs more generally. In 2017, Germany, Netherlands and France agreed to establish an international office to support the FAIR initiative, the GO FAIR International Support and Coordination Office. Other international organisations active in the research data ecosystem, such as CODATA or Research Data Alliance (RDA) also support FAIR implementations by their communities. FAIR principles implementation assessment is being explored by FAIR Data Maturity Model Working Group of RDA, CODATA's strategic Decadal Programme "Data for Planet: Making data work for cross-domain challenges" mentions FAIR data principles as a fundamental enabler of data driven science. The Association of European Research Libraries recommends the use of FAIR principles. A 2017 paper by advocates of FAIR data reported that awareness of the FAIR concept was increasing among various researchers and institutes, but also, understanding of the concept was becoming confused as different people apply their own differing perspectives to it. Guides on implementing FAIR data practices state that the cost of a data management plan in compliance with FAIR data practices should be 5% of the total research budget. In 2019 the Global Indigenous Data Alliance (GIDA) released the CARE Principles for Indigenous Data Governance as a complementary guide. The CARE principles extend principles outlined in FAIR data to include Collective benefit, Authority to control, Responsibility, and Ethics to ensure data guidelines address historical contexts and power differentials. The CARE Principles for Indigenous Data Governance were drafted at the International Data Week and Research Data Alliance Plenary co-hosted event, "Indigenous Data Sovereignty Principles for the Governance of Indigenous Data Workshop", held 8 November 2018, in Gaborone, Botswana. The lack of information on how to implement the guidelines have led to inconsistent interpretations of them. In January 2020, representatives of nine groups of universities around the world produced the Sorbonne declaration on research data rights, which included a commitment to FAIR data, and called on governments to provide support to enable it. In 2021, researchers identified the FAIR principles as a conceptual component of data catalog software tools, with the other components being metadata management, business context and data responsibility roles. In April 2022, Matthias Scheffler and colleagues argued in Nature that FAIR principles are "a must" so that data mining and artificial intelligence can extract useful scientific information from the data. There have been moves in the geosciences to establish FAIR data by use of decimal georeferencing However, making data (and research outcomes) FAIR is a challenging task, and it is challenging to assess the FAIRness. In 2020, the FAIR Data Maturity Model Working Group published a set of guidelines for assessing "FAIRness".
Read more →
Knuth–Plass line-breaking algorithm

The Knuth–Plass algorithm is a line-breaking algorithm designed for use in Donald Knuth's typesetting program TeX. It integrates the problems of text justification and hyphenation into a single algorithm by using a discrete dynamic programming method to minimize a loss function that attempts to quantify the aesthetic qualities desired in the finished output. The algorithm works by dividing the text into a stream of three kinds of objects: boxes, which are non-resizable chunks of content, glue, which are flexible, resizeable elements, and penalties, which represent places where breaking is undesirable (or, if negative, desirable). The loss function, known as "badness", is defined in terms of the deformation of the glue elements, and any extra penalties incurred through line breaking. Making hyphenation decisions follows naturally from the algorithm, but the choice of possible hyphenation points within words, and optionally their preference weighting, must be performed first, and that information inserted into the text stream in advance. Knuth and Plass' original algorithm does not include page breaking, but may be modified to interface with a pagination algorithm, such as the algorithm designed by Plass in his PhD thesis. Typically, the cost function for this technique should be modified so that it does not count the space left on the final line of a paragraph; this modification allows a paragraph to end in the middle of a line without penalty. The same technique can also be extended to take into account other factors such as the number of lines or costs for hyphenating long words. == Computational complexity == A naive brute-force exhaustive search for the minimum badness by trying every possible combination of breakpoints would take an impractical O ( 2 n ) {\displaystyle O(2^{n})} time. The classic Knuth-Plass dynamic programming approach to solving the minimization problem is a worst-case O ( n 2 ) {\displaystyle O(n^{2})} algorithm but usually runs much faster, in close to linear time. Solving for the Knuth-Plass optimum can be shown to be a special case of the convex least-weight subsequence problem, which can be solved in O ( n ) {\displaystyle O(n)} time. Methods to do this include the SMAWK algorithm. == Simple example of minimum raggedness metric == For the input text AAA BB CC DDDDD with line width 6, a greedy algorithm that puts as many words on a line as possible while preserving order before moving to the next line, would produce: ------ Line width: 6 AAA BB Remaining space: 0 CC Remaining space: 4 DDDDD Remaining space: 1 The sum of squared space left over by this method is 0 2 + 4 2 + 1 2 = 17 {\displaystyle 0^{2}+4^{2}+1^{2}=17} . However, the optimal solution achieves the smaller sum 3 2 + 1 2 + 1 2 = 11 {\displaystyle 3^{2}+1^{2}+1^{2}=11} : ------ Line width: 6 AAA Remaining space: 3 BB CC Remaining space: 1 DDDDD Remaining space: 1 The difference here is that the first line is broken before BB instead of after it, yielding a better right margin and a lower cost 11.
Read more →
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 →
List of assembly software and tools

This is a list of assembly software and tools, including software used for assembly language programming, machine code generation, disassembly, debugging, binary analysis, reverse engineering, and instruction-set simulation. == Assemblers and machine-code generators == == Disassemblers and binary-analysis tools == == Debuggers with assembly-level features == == Educational IDEs, simulators and emulators == == Portable and intermediate assembly-like languages == == Assembly language families == Assembly language is not a single programming language, but a family of low-level languages associated with particular instruction set architectures and processor families. Examples include:
Read more →
Library history

Library history is a subdiscipline within library science and library and information science focusing on the history of libraries and their role in societies and cultures. Some see the field as a subset of information history. Library history is an academic discipline and should not be confused with its object of study (history of libraries): the discipline is much younger than the libraries it studies. Library history begins in ancient societies through contemporary issues facing libraries today. Topics include recording mediums, cataloguing systems, scholars, scribes, library supporters and librarians. == Earliest libraries == The earliest records of a library institution as it is presently understood can be dated back to around 5,000 years ago in the Southwest Asian regions of the world. One of the oldest libraries found is that of the ancient library at Ebla (circa 2500 BCE) in present-day Syria. In the 1970s, the excavation at Ebla's library unearthed over 20,000 clay tablets written in cuneiform script. === Library in Mesopotamia === The Assyrian King Assurbanipal created one of the greatest libraries in Nineveh in the seventh century BCE. The collection consisted of over 30,000 tablets written in a variety of languages. The collection was cataloged both by the shape of the tablet and by the subject of the content. The library had separate rooms for the different topics: government, history, law, astronomy, geography, and so on. The tablets also contained myths, hymns, and even jokes. Assurbanipal would send scribes to visit every corner of his kingdom to copy the content of other libraries. His library contained many of the most important literary works of the day, including the epic of Gilgamesh. Assurbanipal's Royal Library also had one of the first library catalogs. Unfortunately, Nineveh was eventually destroyed, and the library was lost in a fire. === Libraries in Ancient Greece === The Greek government was the first to sponsor public libraries. By 500 BCE both Athens and Samos had begun creating libraries for the public, though as most of the population was illiterate these spaces were serving a small, educated portion of the community. Athens developed a city archive at the Metroon in 405 BCE, where documents were stored in sealed jars. These would have saved the documents, but they would have been difficult to consult regularly. In Paros, around the same time, contracts were placed in the temple for safe keeping, and a book curse was placed for extra protection. === Library of Alexandria === The Library at Alexandria, Egypt, was renowned in the third century BCE while kings Ptolemy I Soter and Ptolemy II Philadelphus reigned. The library included a museum, garden, meeting areas and of course reading rooms. The Great Library, as it is known, was one of many in Alexandria. From its inception around the second century BCE, Alexandria was a well-known center for learning. It earned renown as the intellectual capital of the Western world up through the third century CE. The librarians at Alexandria collected, copied, and organized scrolls from across the known world. According to a primary source, every ship that came to Alexandria was required to hand over their books to be copied, and the copies would be returned to the owner, the library keeping the original. The Library of Alexandria was damaged by various disasters over time, including fire, invasion, and earthquake. Scholars believe the collection slowly diminished over time due to theft and efforts to remove it ahead of invading armies. While there are popular stories about how the library was ultimately destroyed, most of these are more myth than fact. === Libraries in Rome === Julius Caesar and his successor Augustus were the first to establish public libraries in ancient Rome, including the library of Apollo on the Palatine Hill. Several emperors followed suit over the next four centuries, including Hadrian, Tiberius, and Vespasian. Roman aristocrats also had personal libraries, which usually contained works in both Greek and Latin. A valuable example of this has been found at Herculaneum near Pompeii. Papyrus manuscripts in Herculaneum's Villa of the Papyri were encased in ash after the eruption of Vesuvius in 79 CE. Modern archaeology is now able to scan these artifacts and discern their contents, including many writings from Philodemus. The average Roman would not have been familiar with books beyond what they might hear read aloud in the forum. Public figures would pay for particular passages to be read aloud to the public from the steps of a public library. === Libraries in the Middle Ages === In the European Middle Ages, libraries began to become more prevalent, despite a widespread reduction in new writing beyond religious themes. Most libraries were initially connected to monasteries or religious institutions. Scriptoriums copied Christian religious texts to share with other religious centers or to be read aloud to their own parishioners. The Holy Roman Emperor Charlemagne (r. 786-814) had a large impact on the advancement of written culture in the Medieval Christian world, acquiring as many written works as he could, and employing many scribes to copy and recirculate vernacular versions of religious works. Most of the text held in small personal libraries was still religious in nature. == Early modern libraries == === Libraries of the Renaissance === During the Renaissance era the merchant middle class grew, and more people found benefits in education. They relied on libraries as a place to study and gain knowledge. Libraries provided a valuable resource, enriching the culture of those who were educated. Universities that had been started in the Middle Ages, founded their own libraries. Books in these libraries could not be borrowed from these libraries and were generally chained to the shelves to prevent theft. As more of the population became literate, new ideas like Humanism and Natural Law spawned an increase personal libraries, although they remained small. Gutenberg's invention of the printing press in 1456 opened the door to the modern era for libraries. == Oldest working libraries == According to the German librarian Michael Knoche, it is not possible to determine which library is the “oldest”: "Precise year dates are a construct, especially in the case of very old libraries. When a collection of books deserves to be called a library depends very much on the point of view of the observer." Various libraries are referred to as the “oldest”: The library founded in the 6th century of the Saint Catherine's Monastery in Sinai is "reputedly the oldest continuously run library in existence today", according to the Library of Congress. Its collection of religious and secular manuscripts is ranging from Bibles, liturgies and prayer books to legal documents such as deeds, court cases and fatwahs (legal opinions). The Al Qarawiyyin Library was founded in 859 by Fatima al-Fihri and is often regarded as the oldest working library in the world. It is in Fez, Morocco and is part of the oldest continually operating university in the world, the University of al-Qarawiyyin. The library houses approximately 4,000 ancient Islamic manuscripts. These manuscripts include 9th century Qurans and the oldest known accounts of the Islamic prophet Muhammed. The Malatestiana Library (Italian: Biblioteca Malatestiana) is a public library in the city of Cesena in northern Italy. Opened in 1454 it is significant for being the first civic library in Europe open to the general public. == Library history reports and writings of the early 19th and 20th century == In the early 19th and 20th century, representative titles were created reporting library history in the United States and the United Kingdom. American titles include Public Libraries in the United States of America, Their History, Condition, and Management (1876), Memorial History of Boston (1881) by Justin Winsor, Public Libraries in America (1894) by William I. Fletcher, and History of the New York Public Library (1923) by Henry M. Lydenberg. British titles include Old English Libraries (1911) by Earnest A. Savage and The Chained Library: A Survey of Four Centuries in the Evolution of the English Library by Burnett Hillman Streeter. In the beginning of the 20th century, library historians began applying scientific research methodologies to examine the library as a social agency. Two works that demonstrate this argument are Geschichte der Bibliotheken (1925) by Alfred Hessel and the Library Quarterly article from 1931, “The Sociological Beginnings of the Library Movement in America” by Arnold Borden. With the establishment of library schools, master's theses and doctoral dissertations represented the shift in serious research regarding libraries and library history. Two published doctoral dissertations that mark this trend are Foundations of the Public Library: The Origins of the American Public Library Movement in Ne
Read more →
Algorithmic transparency

Algorithmic transparency is the principle that the factors that influence the decisions made by algorithms should be visible, or transparent, to the people who use, regulate, and are affected by systems that employ those algorithms. Although the phrase was coined in 2016 by Nicholas Diakopoulos and Michael Koliska about the role of algorithms in deciding the content of digital journalism services, the underlying principle dates back to the 1970s and the rise of automated systems for scoring consumer credit. The phrases "algorithmic transparency" and "algorithmic accountability" are sometimes used interchangeably – especially since they were coined by the same people – but they have subtly different meanings. Specifically, "algorithmic transparency" states that the inputs to the algorithm and the algorithm's use itself must be known, but they need not be fair. "Algorithmic accountability" implies that the organizations that use algorithms must be accountable for the decisions made by those algorithms, even though the decisions are being made by a machine, and not by a human being. Current research around algorithmic transparency interested in both societal effects of accessing remote services running algorithms, as well as mathematical and computer science approaches that can be used to achieve algorithmic transparency. In the United States, the Federal Trade Commission's Bureau of Consumer Protection studies how algorithms are used by consumers by conducting its own research on algorithmic transparency and by funding external research. In the European Union, the data protection laws that came into effect in May 2018 include a "right to explanation" of decisions made by algorithms, though it is unclear what this means. Furthermore, the European Union founded The European Center for Algorithmic Transparency (ECAT).
Read more →
EdgeRank

EdgeRank is the name commonly given to the algorithm that Facebook uses to determine what articles should be displayed in a user's News Feed. As of 2011, Facebook has stopped using the EdgeRank system and uses a machine learning algorithm that, as of 2013, takes more than 100,000 factors into account. EdgeRank was developed and implemented by Serkan Piantino. == Formula and factors == In 2010, a simplified version of the EdgeRank algorithm was presented as: ∑ e d g e s e u e w e d e {\displaystyle \sum _{\mathrm {edges\,} e}u_{e}w_{e}d_{e}} where: u e {\displaystyle u_{e}} is user affinity. w e {\displaystyle w_{e}} is how the content is weighted. d e {\displaystyle d_{e}} is a time-based decay parameter. User Affinity: The User Affinity part of the algorithm in Facebook's EdgeRank looks at the relationship and proximity of the user and the content (post/status update). Content Weight: What action was taken by the user on the content. Time-Based Decay Parameter: New or old. Newer posts tend to hold a higher place than older posts. Some of the methods that Facebook uses to adjust the parameters are proprietary and not available to the public. A study has shown that it is possible to hypothesize a disadvantage of the "like" reaction and advantages of other interactions (e.g., the "haha" reaction or "comments") in content algorithmic ranking on Facebook. The "like" button can decrease the organic reach as a "brake effect of viral reach". The "haha" reaction, "comments" and the "love" reaction could achieve the highest increase in total organic reach. == Impact == EdgeRank and its successors have a broad impact on what users actually see out of what they ostensibly follow: for instance, the selection can produce a filter bubble (if users are exposed to updates which confirm their opinions etc.) or alter people's mood (if users are shown a disproportionate amount of positive or negative updates). As a result, for Facebook pages, the typical engagement rate is less than 1% (or less than 0.1% for the bigger ones), and organic reach 10% or less for most non-profits. As a consequence, for pages, it may be nearly impossible to reach any significant audience without paying to promote their content.
Read more →
Neural scaling law

In machine learning, a neural scaling law is an empirical scaling law that describes how neural network performance changes as key factors are scaled up or down. These factors typically include the number of parameters, training dataset size, and training cost. Some models also exhibit performance gains by scaling inference through increased test-time compute (TTC), extending neural scaling laws beyond training to the deployment phase. == Introduction == In general, a deep learning model can be characterized by four parameters: model size, training dataset size, training cost, and the post-training error rate (e.g., the test set error rate). Each of these variables can be defined as a real number, usually written as N , D , C , L {\displaystyle N,D,C,L} (respectively: parameter count, dataset size, computing cost, and loss). A neural scaling law is a theoretical or empirical statistical law between these parameters. There are also other parameters with other scaling laws. === Size of the model === In most cases, the model's size is simply the number of parameters. However, one complication arises with the use of sparse models, such as mixture-of-expert models. With sparse models, during inference, only a fraction of their parameters are used. In comparison, most other kinds of neural networks, such as transformer models, always use all their parameters during inference. === Size of the training dataset === The size of the training dataset is usually quantified by the number of data points within it. Larger training datasets are typically preferred, as they provide a richer and more diverse source of information from which the model can learn. This can lead to improved generalization performance when the model is applied to new, unseen data. However, increasing the size of the training dataset also increases the computational resources and time required for model training. With the "pretrain, then finetune" method used for most large language models, there are two kinds of training dataset: the pretraining dataset and the finetuning dataset. Their sizes have different effects on model performance. Generally, the finetuning dataset is less than 1% the size of pretraining dataset. In some cases, a small amount of high quality data suffices for finetuning, and more data does not necessarily improve performance. Many scaling laws, due to their inherent diminishing returns nature, value data based on a submodular set function which was shown in a paper on this topic. === Cost of training === Training cost is typically measured in terms of time (how long it takes to train the model) and computational resources (how much processing power and memory are required). It is important to note that the cost of training can be significantly reduced with efficient training algorithms, optimized software libraries, and parallel computing on specialized hardware such as GPUs or TPUs. The cost of training a neural network model is a function of several factors, including model size, training dataset size, the training algorithm complexity, and the computational resources available. In particular, doubling the training dataset size does not necessarily double the cost of training, because one may train the model for several times over the same dataset (each being an "epoch"). === Performance === The performance of a neural network model is evaluated based on its ability to accurately predict the output given some input data. Common metrics for evaluating model performance include: Negative log-likelihood per token (logarithm of perplexity) for language modeling; Accuracy, precision, recall, and F1 score for classification tasks; Mean squared error (MSE) or mean absolute error (MAE) for regression tasks; Elo rating in a competition against other models, such as gameplay or preference by a human judge. Performance can be improved by using more data, larger models, different training algorithms, regularizing the model to prevent overfitting, and early stopping using a validation set. When the performance is a number bounded within the range of [ 0 , 1 ] {\displaystyle [0,1]} , such as accuracy, precision, etc., it often scales as a sigmoid function of cost, as seen in the figures. == Examples == === (Hestness, Narang, et al, 2017) === The 2017 paper is a common reference point for neural scaling laws fitted by statistical analysis on experimental data. Previous works before the 2000s, as cited in the paper, were either theoretical or orders of magnitude smaller in scale. Whereas previous works generally found the scaling exponent to scale like L ∝ D − α {\displaystyle L\propto D^{-\alpha }} , with α ∈ { 0.5 , 1 , 2 } {\displaystyle \alpha \in \{0.5,1,2\}} , the paper found that α ∈ [ 0.07 , 0.35 ] {\displaystyle \alpha \in [0.07,0.35]} . Of the factors they varied, only task can change the exponent α {\displaystyle \alpha } . Changing the architecture optimizers, regularizers, and loss functions, would only change the proportionality factor, not the exponent. For example, for the same task, one architecture might have L = 1000 D − 0.3 {\displaystyle L=1000D^{-0.3}} while another might have L = 500 D − 0.3 {\displaystyle L=500D^{-0.3}} . They also found that for a given architecture, the number of parameters necessary to reach lowest levels of loss, given a fixed dataset size, grows like N ∝ D β {\displaystyle N\propto D^{\beta }} for another exponent β {\displaystyle \beta } . They studied machine translation with LSTM ( α ∼ 0.13 {\displaystyle \alpha \sim 0.13} ), generative language modelling with LSTM ( α ∈ [ 0.06 , 0.09 ] , β ≈ 0.7 {\displaystyle \alpha \in [0.06,0.09],\beta \approx 0.7} ), ImageNet classification with ResNet ( α ∈ [ 0.3 , 0.5 ] , β ≈ 0.6 {\displaystyle \alpha \in [0.3,0.5],\beta \approx 0.6} ), and speech recognition with two hybrid (LSTMs complemented by either CNNs or an attention decoder) architectures ( α ≈ 0.3 {\displaystyle \alpha \approx 0.3} ). === (Henighan, Kaplan, et al, 2020) === A 2020 analysis studied statistical relations between C , N , D , L {\displaystyle C,N,D,L} over a wide range of values and found similar scaling laws, over the range of N ∈ [ 10 3 , 10 9 ] {\displaystyle N\in [10^{3},10^{9}]} , C ∈ [ 10 12 , 10 21 ] {\displaystyle C\in [10^{12},10^{21}]} , and over multiple modalities (text, video, image, text to image, etc.). In particular, the scaling laws it found are (Table 1 of ): For each modality, they fixed one of the two C , N {\displaystyle C,N} , and varying the other one ( D {\displaystyle D} is varied along using D = C / 6 N {\displaystyle D=C/6N} ), the achievable test loss satisfies L = L 0 + ( x 0 x ) α {\displaystyle L=L_{0}+\left({\frac {x_{0}}{x}}\right)^{\alpha }} where x {\displaystyle x} is the varied variable, and L 0 , x 0 , α {\displaystyle L_{0},x_{0},\alpha } are parameters to be found by statistical fitting. The parameter α {\displaystyle \alpha } is the most important one. When N {\displaystyle N} is the varied variable, α {\displaystyle \alpha } ranges from 0.037 {\displaystyle 0.037} to 0.24 {\displaystyle 0.24} depending on the model modality. This corresponds to the α = 0.34 {\displaystyle \alpha =0.34} from the Chinchilla scaling paper. When C {\displaystyle C} is the varied variable, α {\displaystyle \alpha } ranges from 0.048 {\displaystyle 0.048} to 0.19 {\displaystyle 0.19} depending on the model modality. This corresponds to the β = 0.28 {\displaystyle \beta =0.28} from the Chinchilla scaling paper. Given fixed computing budget, optimal model parameter count is consistently around N o p t ( C ) = ( C 5 × 10 − 12 petaFLOP-day ) 0.7 = 9.0 × 10 − 7 C 0.7 {\displaystyle N_{opt}(C)=\left({\frac {C}{5\times 10^{-12}{\text{petaFLOP-day}}}}\right)^{0.7}=9.0\times 10^{-7}C^{0.7}} The parameter 9.0 × 10 − 7 {\displaystyle 9.0\times 10^{-7}} varies by a factor of up to 10 for different modalities. The exponent parameter 0.7 {\displaystyle 0.7} varies from 0.64 {\displaystyle 0.64} to 0.75 {\displaystyle 0.75} for different modalities. This exponent corresponds to the ≈ 0.5 {\displaystyle \approx 0.5} from the Chinchilla scaling paper. It's "strongly suggested" (but not statistically checked) that D o p t ( C ) ∝ N o p t ( C ) 0.4 ∝ C 0.28 {\displaystyle D_{opt}(C)\propto N_{opt}(C)^{0.4}\propto C^{0.28}} . This exponent corresponds to the ≈ 0.5 {\displaystyle \approx 0.5} from the Chinchilla scaling paper. The scaling law of L = L 0 + ( C 0 / C ) 0.048 {\displaystyle L=L_{0}+(C_{0}/C)^{0.048}} was confirmed during the training of GPT-3 (Figure 3.1 ). === Chinchilla scaling (Hoffmann, et al, 2022) === One particular scaling law ("Chinchilla scaling") states that, for a large language model (LLM) autoregressively trained for one epoch, with a cosine learning rate schedule, we have: { C = C 0 N D L = A N α + B D β + L 0 {\displaystyle {\begin{cases}C=C_{0}ND\\L={\frac {A}{N^{\alpha }}}+{\frac {B}{D^{\beta }}}+L_{0}\end{cases}}} where the variables are C {\displaystyle C} is the cost o
Read more →
BioCreative

BioCreAtIvE (A critical assessment of text mining methods in molecular biology) consists in a community-wide effort for evaluating information extraction and text mining developments in the biological domain. It was preceded by the Knowledge Discovery and Data Mining (KDD) Challenge Cup for detection of gene mentions. == Community Challenges == === First edition (2004-2005) === Three main tasks were posed at the first BioCreAtIvE challenge: the entity extraction task, the gene name normalization task, and the functional annotation of gene products task. The data sets produced by this contest serve as a Gold Standard training and test set to evaluate and train Bio-NER tools and annotation extraction tools. === Second edition (2006-2007) === The second BioCreAtIvE challenge (2006-2007) had also 3 tasks: detection of gene mentions, extraction of unique idenfiers for genes and extraction information related to physical protein-protein interactions. It counted with participation of 44 teams from 13 countries. === Third edition (2011-2012) === The third edition of BioCreative included for the first time the InterActive Task (IAT), designed to evaluate the practical usability of text mining tools in real-world biocuration tasks. === Fifth edition (2016) === BioCreative V had 5 different tracks, including an interactive task (IAT) for usability of text mining systems and a track using the BioC format for curating information for BioGRID.
Read more →
Knowledge organization

Knowledge organization (KO), organization of knowledge, organization of information, or information organization is an intellectual discipline concerned with activities such as document description, indexing, and classification that serve to provide systems of representation and order for knowledge and information objects. According to The Organization of Information by Joudrey and Taylor, information organization: examines the activities carried out and tools used by people who work in places that accumulate information resources (e.g., books, maps, documents, datasets, images) for the use of humankind, both immediately and for posterity. It discusses the processes that are in place to make resources findable, whether someone is searching for a single known item or is browsing through hundreds of resources just hoping to discover something useful. Information organization supports a myriad of information-seeking scenarios. Issues related to knowledge sharing can be said to have been an important part of knowledge management for a long time. Knowledge sharing has received a lot of attention in research and business practice both within and outside organizations and its different levels. Sharing knowledge is not only about giving it to others, but it also includes searching, locating, and absorbing knowledge. Unawareness of the employees' work and duties tends to provoke the repetition of mistakes, the waste of resources, and duplication of the same projects. Motivating co-workers to share their knowledge is called knowledge enabling. It leads to trust among individuals and encourages a more open and proactive relationship that grants the exchange of information easily. Knowledge sharing is part of the three-phase knowledge management process which is a continuous process model. The three parts are knowledge creation, knowledge implementation, and knowledge sharing. The process is continuous, which is why the parts cannot be fully separated. Knowledge creation is the consequence of individuals' minds, interactions, and activities. Developing new ideas and arrangements alludes to the process of knowledge creation. Using the knowledge which is present at the company in the most effective manner stands for the implementation of knowledge. Knowledge sharing, the most essential part of the process for our topic, takes place when two or more people benefit by learning from each other. Traditional human-based approaches performed by librarians, archivists, and subject specialists are increasingly challenged by computational (big data) algorithmic techniques. KO as a field of study is concerned with the nature and quality of such knowledge-organizing processes (KOP) (such as taxonomy and ontology) as well as the resulting knowledge organizing systems (KOS). == Theoretical approaches == === Traditional approaches === Among the major figures in the history of KO are Melvil Dewey (1851–1931) and Henry Bliss (1870–1955). Dewey's goal was an efficient way to manage library collections; not an optimal system to support users of libraries. His system was meant to be used in many libraries as a standardized way to manage collections. The first version of this system was created in 1876. An important characteristic in Henry Bliss' (and many contemporary thinkers of KO) was that the sciences tend to reflect the order of Nature and that library classification should reflect the order of knowledge as uncovered by science: The implication is that librarians, in order to classify books, should know about scientific developments. This should also be reflected in their education: Again from the standpoint of the higher education of librarians, the teaching of systems of classification ... would be perhaps better conducted by including courses in the systematic encyclopedia and methodology of all the sciences, that is to say, outlines which try to summarize the most recent results in the relation to one another in which they are now studied together. ... (Ernest Cushing Richardson, quoted from Bliss, 1935, p. 2) Among the other principles, which may be attributed to the traditional approach to KO are: Principle of controlled vocabulary Cutter's rule about specificity Hulme's principle of literary warrant (1911) Principle of organizing from the general to the specific Today, after more than 100 years of research and development in LIS, the "traditional" approach still has a strong position in KO and in many ways its principles still dominate. === Facet analytic approaches === The date of the foundation of this approach may be chosen as the publication of S. R. Ranganathan's colon classification in 1933. The approach has been further developed by, in particular, the British Classification Research Group. The best way to explain this approach is probably to explain its analytico-synthetic methodology. The meaning of the term "analysis" is: breaking down each subject into its basic concepts. The meaning of the term synthesis is: combining the relevant units and concepts to describe the subject matter of the information package in hand. Given subjects (as they appear in, for example, book titles) are first analyzed into a few common categories, which are termed "facets". Ranganathan proposed his PMEST formula: Personality, Matter, Energy, Space and Time: Personality is the distinguishing characteristic of a subject. Matter is the physical material of which a subject may be composed. Energy is any action that occurs with respect to the subject. Space is the geographic component of the location of a subject. Time is the period associated with a subject. === The information retrieval tradition (IR) === Important in the IR-tradition have been, among others, the Cranfield experiments, which were founded in the 1950s, and the TREC experiments (Text Retrieval Conferences) starting in 1992. It was the Cranfield experiments, which introduced the measures "recall" and "precision" as evaluation criteria for systems efficiency. The Cranfield experiments found that classification systems like UDC and facet-analytic systems were less efficient compared to free-text searches or low level indexing systems ("UNITERM"). The Cranfield I test found, according to Ellis (1996, 3–6) the following results: Although these results have been criticized and questioned, the IR-tradition became much more influential while library classification research lost influence. The dominant trend has been to regard only statistical averages. What has largely been neglected is to ask: Are there certain kinds of questions in relation to which other kinds of representation, for example, controlled vocabularies, may improve recall and precision? === User-oriented and cognitive views === The best way to define this approach is probably by method: Systems based upon user-oriented approaches must specify how the design of a system is made on the basis of empirical studies of users. User studies demonstrated very early that users prefer verbal search systems as opposed to systems based on classification notations. This is one example of a principle derived from empirical studies of users. Adherents of classification notations may, of course, still have an argument: That notations are well-defined and that users may miss important information by not considering them. Folksonomies is a recent kind of KO based on users' rather than on librarians' or subject specialists' indexing. === Bibliometric approaches === These approaches are primarily based on using bibliographical references to organize networks of papers, mainly by bibliographic coupling (introduced by Kessler 1963) or co-citation analysis ( independently suggested by Marshakova 1973 and Small 1973). In recent years it has become a popular activity to construe bibliometric maps as structures of research fields. Two considerations are important in considering bibliometric approaches to KO: The level of indexing depth is partly determined by the number of terms assigned to each document. In citation indexing this corresponds to the number of references in a given paper. On the average, scientific papers contain 10–15 references, which provide quite a high level of depth. The references, which function as access points, are provided by the highest subject-expertise: The experts writing in the leading journals. This expertise is much higher than that which library catalogs or bibliographical databases typically are able to draw on. === The domain analytic approach === Domain analysis is a sociological-epistemological standpoint that advocates that the indexing of a given document should reflect the needs of a given group of users or a given ideal purpose. In other words, any description or representation of a given document is more or less suited to the fulfillment of certain tasks. A description is never objective or neutral, and the goal is not to standardize descriptions or make one description once and for all for different target groups. The develo
Read more →
Manufacture Modules Technologies

Manufacture Modules Technologies Sarl (MMT) is a Swiss company established in Geneva in 2015 which originally specialised in the development and commercialization of "Horological Smartwatch modules", firmware, apps and cloud. Located at Geneva's Skylab high-tech hub, it expanded into the development and manufacturing of "E-Straps" operated with a mobile application. Philippe Fraboulet is the CEO. == History == In June 2015, Fullpower Technologies and Union Horlogère Suisse (Swiss Watchmakers Corporation) formed MMT as a joint venture, which then launched the MotionX Horological Smartwatch Open Platform for the Swiss watch industry. The initial licensees were Frederique Constant, Alpina and Mondaine, brands owned by Union Horlogère Suisse. Fullpower created and managed the circuit design, firmware, smartphone applications (including sleep activity), as well as the cloud Infrastructure. MMT managed the Swiss watch movement development and production as well as licensing and support. In July 2016, Union Horlogere Holding and MMT were spun-out of the Frédérique Constant Group. Fullpower Technologies' 19.99% share was acquired by Union Horlogere Holding BV, giving it 100% of MMT's shares. == Business == The company offers firmware, a cloud, manufacturing, service and over-the-air facilities for upgrades. The company also offers its own apps, which bear the label “Swiss Made software”.
Read more →
Semantic decomposition (natural language processing)

A semantic decomposition is an algorithm that breaks down the meanings of phrases or concepts into less complex concepts. The result of a semantic decomposition is a representation of meaning. This representation can be used for tasks, such as those related to artificial intelligence or machine learning. Semantic decomposition is common in natural language processing applications. The basic idea of a semantic decomposition is taken from the learning skills of adult humans, where words are explained using other words. It is based on Meaning-text theory. Meaning-text theory is used as a theoretical linguistic framework to describe the meaning of concepts with other concepts. == Background == Given that an AI does not inherently have language, it is unable to think about the meanings behind the words of a language. An artificial notion of meaning needs to be created for a strong AI to emerge. Creating an artificial representation of meaning requires the analysis of what meaning is. Many terms are associated with meaning, including semantics, pragmatics, knowledge and understanding or word sense. Each term describes a particular aspect of meaning, and contributes to a multitude of theories explaining what meaning is. These theories need to be analyzed further to develop an artificial notion of meaning best fit for our current state of knowledge. == Graph representations == Representing meaning as a graph is one of the two ways that both an AI cognition and a linguistic researcher think about meaning (connectionist view). Logicians utilize a formal representation of meaning to build upon the idea of symbolic representation, whereas description logics describe languages and the meaning of symbols. This contention between 'neat' and 'scruffy' techniques has been discussed since the 1970s. Research has so far identified semantic measures and with that word-sense disambiguation (WSD) - the differentiation of meaning of words - as the main problem of language understanding. As an AI-complete environment, WSD is a core problem of natural language understanding. AI approaches that use knowledge-given reasoning creates a notion of meaning combining the state of the art knowledge of natural meaning with the symbolic and connectionist formalization of meaning for AI. The abstract approach is shown in Figure. First, a connectionist knowledge representation is created as a semantic network consisting of concepts and their relations to serve as the basis for the representation of meaning. This graph is built out of different knowledge sources like WordNet, Wiktionary, and BabelNET. The graph is created by lexical decomposition that recursively breaks each concept semantically down into a set of semantic primes. The primes are taken from the theory of Natural Semantic Metalanguage, which has been analyzed for usefulness in formal languages. Upon this graph marker passing is used to create the dynamic part of meaning representing thoughts. The marker passing algorithm, where symbolic information is passed along relations form one concept to another, uses node and edge interpretation to guide its markers. The node and edge interpretation model is the symbolic influence of certain concepts. Future work uses the created representation of meaning to build heuristics and evaluate them through capability matching and agent planning, chatbots or other applications of natural language understanding.
Read more →
Subject (documents)

In library and information science documents (such as books, articles and pictures) are classified and searched by subject – as well as by other attributes such as author, genre and document type. This makes "subject" a fundamental term in this field. Library and information specialists assign subject labels to documents to make them findable. There are many ways to do this and in general there is not always consensus about which subject should be assigned to a given document. To optimize subject indexing and searching, we need to have a deeper understanding of what a subject is. The question: "what is to be understood by the statement 'document A belongs to subject category X'?" has been debated in the field for more than 100 years (see below) == Theoretical view == === Charles Ammi Cutter (1837–1903) === For Cutter the stability of subjects depends on a social process in which their meaning is stabilized in a name or a designation. A subject "referred [...] to those intellections [...] that had received a name that itself represented a distinct consensus in usage" (Miksa, 1983a, p. 60) and: the "systematic structure of established subjects" is "resident in the public realm" (Miksa, 1983a, p. 69); "[s]ubjects are by their very nature locations in a classificatory structure of publicly accumulated knowledge (Miksa, 1983a, p. 61). Bernd Frohmann adds: "The stability of the public realm in turn relies upon natural and objective mental structures which, with proper education, govern a natural progression from particular to general concepts. Since for Cutter, mind, society, and SKO [Systems of Knowledge Organization] stand one behind the other, each supporting each, all manifesting the same structure, his discursive construction of subjects invites connections with discourses of mind, education, and society. The Dewey Decimal Classification (DDC), by contrast, severs those connections. Melvil Dewey emphasized more than once that his system maps no structure beyond its own; there is neither a "transcendental deduction" of its categories nor any reference to Cutter's objective structure of social consensus. It is content-free: Dewey disdained any philosophical excogitation of the meaning of his class symbols, leaving the job of finding verbal equivalents to others. His innovation and the essence of the system lay in the notation. The DDC is a poorly semiotic system of expanding nests of ten digits, lacking any referent beyond itself. In it, a subject is wholly constituted in terms of its position in the system. The essential characteristic of a subject is a class symbol which refers only to other symbols. Its verbal equivalent is accidental, a merely pragmatic characteristic... .... The conflict of interpretations over "subjects" became explicit in the battles between "bibliography" (an approach to subjects having much in common with Cutter's) and Dewey's "close classification". William Fletcher spoke for the scholarly bibliographer.... Fletcher's "subjects", like Cutter's, referred to the categories of a fantasized, stable social order, whereas Dewey's subjects were elements of a semiological system of standardized, techno-bureaucratic administrative software for the library in its corporate, rather than high culture, incarnation". (Frohmann, 1994, 112–113). Cutter's early view on what a subject is, is probably wiser than most understandings that dominated the 20th century – and also the understanding reflected in the ISO-standard quoted below. The early statements quoted by Frohmann indicate that subjects are somehow shaped in social processes. When that is said, it should be added that they are not particularly detailed or clear. We only get a vague idea of the social nature of subjects. === S. R. Ranganathan (1892–1972) === A classification system with an explicit theoretical foundation is Ranganathan's Colon Classification. Ranganathan provided an explicit definition of the concept of "subject": Subject – an organized body of ideas, whose extension and intension are likely to fall coherently within the field of interests and comfortably within the intellectual competence and the field of inevitable specialization of a normal person. A related definition is given by one of Ranganathan's students: A subject is an organized and systematized body of ideas. It may consist of one idea or a combination of several... Ranganathan's definition of "subject" is strongly influenced by his Colon Classification system. The colon system is based on the combination of single elements from facets to subject designation. This is the reason why the combined nature of subjects are emphasized so strongly. It leads, however, to absurdities such as the claim that gold cannot be a subject (but is alternatively termed "an isolate"). This aspect of the theory has been criticized by Metcalfe (1973, p. 318). Metcalfe's skepticism regarding Ranganathan's theory is formulated in hard words (op. cit., p. 317): "This pseudo-science imposed itself on British disciples from about 1950 on...". It seems unacceptable that Ranganathan defines the word subject in a way that favors his own system. A scientific concept like "subject" should make it possible to compare different ways of establishing access to information. Whether or not subjects are combined or not should be examined once their definition has been given, it should not determined a priori, in the definition. Besides the emphasis on the combined, organizing and systematizing nature of subjects contains Ranganathan's definition of subject the pragmatic demand, that a subject should be determined in a way that suits a normal person's competency or specialization. Again we see a strange kind of wishful thinking mixing a general understanding of a concept with demands put by his own specific system. One thing is what the word subject means, quite another issue is how to provide subject descriptions that fulfill demands such as the specificity of a given information retrieval language which fulfill demands put on the system, such as precision and recall. If researchers too often define terms in ways that favor specific kinds of systems, that are such definitions not useful to provide more general theories about subjects, subject analysis and IR. Among other things are comparative studies of different kinds of systems made difficult. Based on these arguments, as well as additional arguments which have been used in the literature, we may conclude that Ranganathan's definition of the concept "subject" is not suited for scientific use. Like the definition of "subject" given by the ISO-standard for topic maps, may Ranganathan's definition be useful within his own closed system. The purpose of a scientific and scholarly field is, however, to examine the relative fruitfulness of systems such as topic maps and Colon Classification. For such purpose is another understanding of "subject" necessary. === Patrick Wilson (1927–2003) === In his book Wilson (1968) examined – in particular by thought experiments – the suitability of different methods of examining the subject of a document. The methods were: identifying the author's purpose for writing the document, weighing the relative dominance and subordination of different elements in the picture, which the reading imposes on the reader, grouping or count the document's use of concepts and references, construing a set of rules for selecting elements deemed necessary (as opposed to unnecessary) for the work as a whole. Patrick Wilson shows convincingly that each of these methods are insufficient to determine the subject of a document and is led to conclude ( p. 89): "The notion of the subject of a writing is indeterminate..." or, on p. 92 (about what users may expect to find using a particular position in a library classification system): "For nothing definite can be expected of the things found at any given position". In connection to the last quote has Wilson an interesting footnote in which he writes that authors of documents often use terms in ambiguous ways ("hostility" is used as an example). Even if the librarian could personally develop a very precise understanding of a concept, he would be unable to use it in his classification, because none of the documents use the term in the same precise way. Based on this argumentation is Wilson led to conclude: "If people write on what are for them ill-defined phenomena, a correct description of their subjects must reflect the ill-definedness". Wilson's concept of subject was discussed by Hjørland (1992) who found that it is problematic to give up the precise understanding of such a basic term in LIS. Wilson's arguments led him to an agnostic position which Hjørland found unacceptable and unnecessary. Concerning the authors' use of ambiguous terms, the role of the subject analysis is to determine which documents would be fruitful for users to identify whether or not the documents use one or another term or whether a given term i
Read more →
Zassenhaus algorithm

In mathematics, the Zassenhaus algorithm is a method to calculate a basis for the intersection and sum of two subspaces of a vector space. It is named after Hans Zassenhaus, but no publication of this algorithm by him is known. It is used in computer algebra systems. == Algorithm == === Input === Let V be a vector space and U, W two finite-dimensional subspaces of V with the following spanning sets: U = ⟨ u 1 , … , u n ⟩ {\displaystyle U=\langle u_{1},\ldots ,u_{n}\rangle } and W = ⟨ w 1 , … , w k ⟩ . {\displaystyle W=\langle w_{1},\ldots ,w_{k}\rangle .} Finally, let B 1 , … , B m {\displaystyle B_{1},\ldots ,B_{m}} be linearly independent vectors so that u i {\displaystyle u_{i}} and w i {\displaystyle w_{i}} can be written as u i = ∑ j = 1 m a i , j B j {\displaystyle u_{i}=\sum _{j=1}^{m}a_{i,j}B_{j}} and w i = ∑ j = 1 m b i , j B j . {\displaystyle w_{i}=\sum _{j=1}^{m}b_{i,j}B_{j}.} === Output === The algorithm computes the base of the sum U + W {\displaystyle U+W} and a base of the intersection U ∩ W {\displaystyle U\cap W} . === Algorithm === The algorithm creates the following block matrix of size ( ( n + k ) × ( 2 m ) ) {\displaystyle ((n+k)\times (2m))} : ( a 1 , 1 a 1 , 2 ⋯ a 1 , m a 1 , 1 a 1 , 2 ⋯ a 1 , m ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ a n , 1 a n , 2 ⋯ a n , m a n , 1 a n , 2 ⋯ a n , m b 1 , 1 b 1 , 2 ⋯ b 1 , m 0 0 ⋯ 0 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ b k , 1 b k , 2 ⋯ b k , m 0 0 ⋯ 0 ) {\displaystyle {\begin{pmatrix}a_{1,1}&a_{1,2}&\cdots &a_{1,m}&a_{1,1}&a_{1,2}&\cdots &a_{1,m}\\\vdots &\vdots &&\vdots &\vdots &\vdots &&\vdots \\a_{n,1}&a_{n,2}&\cdots &a_{n,m}&a_{n,1}&a_{n,2}&\cdots &a_{n,m}\\b_{1,1}&b_{1,2}&\cdots &b_{1,m}&0&0&\cdots &0\\\vdots &\vdots &&\vdots &\vdots &\vdots &&\vdots \\b_{k,1}&b_{k,2}&\cdots &b_{k,m}&0&0&\cdots &0\end{pmatrix}}} Using elementary row operations, this matrix is transformed to the row echelon form. Then, it has the following shape: ( c 1 , 1 c 1 , 2 ⋯ c 1 , m ∙ ∙ ⋯ ∙ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ c q , 1 c q , 2 ⋯ c q , m ∙ ∙ ⋯ ∙ 0 0 ⋯ 0 d 1 , 1 d 1 , 2 ⋯ d 1 , m ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ 0 0 ⋯ 0 d ℓ , 1 d ℓ , 2 ⋯ d ℓ , m 0 0 ⋯ 0 0 0 ⋯ 0 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ 0 0 ⋯ 0 0 0 ⋯ 0 ) {\displaystyle {\begin{pmatrix}c_{1,1}&c_{1,2}&\cdots &c_{1,m}&\bullet &\bullet &\cdots &\bullet \\\vdots &\vdots &&\vdots &\vdots &\vdots &&\vdots \\c_{q,1}&c_{q,2}&\cdots &c_{q,m}&\bullet &\bullet &\cdots &\bullet \\0&0&\cdots &0&d_{1,1}&d_{1,2}&\cdots &d_{1,m}\\\vdots &\vdots &&\vdots &\vdots &\vdots &&\vdots \\0&0&\cdots &0&d_{\ell ,1}&d_{\ell ,2}&\cdots &d_{\ell ,m}\\0&0&\cdots &0&0&0&\cdots &0\\\vdots &\vdots &&\vdots &\vdots &\vdots &&\vdots \\0&0&\cdots &0&0&0&\cdots &0\end{pmatrix}}} Here, ∙ {\displaystyle \bullet } stands for arbitrary numbers, and the vectors ( c p , 1 , c p , 2 , … , c p , m ) {\displaystyle (c_{p,1},c_{p,2},\ldots ,c_{p,m})} for every p ∈ { 1 , … , q } {\displaystyle p\in \{1,\ldots ,q\}} and ( d p , 1 , … , d p , m ) {\displaystyle (d_{p,1},\ldots ,d_{p,m})} for every p ∈ { 1 , … , ℓ } {\displaystyle p\in \{1,\ldots ,\ell \}} are nonzero. Then ( y 1 , … , y q ) {\displaystyle (y_{1},\ldots ,y_{q})} with y i := ∑ j = 1 m c i , j B j {\displaystyle y_{i}:=\sum _{j=1}^{m}c_{i,j}B_{j}} is a basis of U + W {\displaystyle U+W} and ( z 1 , … , z ℓ ) {\displaystyle (z_{1},\ldots ,z_{\ell })} with z i := ∑ j = 1 m d i , j B j {\displaystyle z_{i}:=\sum _{j=1}^{m}d_{i,j}B_{j}} is a basis of U ∩ W {\displaystyle U\cap W} . === Proof of correctness === First, we define π 1 : V × V → V , ( a , b ) ↦ a {\displaystyle \pi _{1}:V\times V\to V,(a,b)\mapsto a} to be the projection to the first component. Let H := { ( u , u ) ∣ u ∈ U } + { ( w , 0 ) ∣ w ∈ W } ⊆ V × V . {\displaystyle H:=\{(u,u)\mid u\in U\}+\{(w,0)\mid w\in W\}\subseteq V\times V.} Then π 1 ( H ) = U + W {\displaystyle \pi _{1}(H)=U+W} and H ∩ ( 0 × V ) = 0 × ( U ∩ W ) {\displaystyle H\cap (0\times V)=0\times (U\cap W)} . Also, H ∩ ( 0 × V ) {\displaystyle H\cap (0\times V)} is the kernel of π 1 | H {\displaystyle {\pi _{1}|}_{H}} , the projection restricted to H. Therefore, dim ⁡ ( H ) = dim ⁡ ( U + W ) + dim ⁡ ( U ∩ W ) {\displaystyle \dim(H)=\dim(U+W)+\dim(U\cap W)} . The Zassenhaus algorithm calculates a basis of H. In the first m columns of this matrix, there is a basis y i {\displaystyle y_{i}} of U + W {\displaystyle U+W} . The rows of the form ( 0 , z i ) {\displaystyle (0,z_{i})} (with z i ≠ 0 {\displaystyle z_{i}\neq 0} ) are obviously in H ∩ ( 0 × V ) {\displaystyle H\cap (0\times V)} . Because the matrix is in row echelon form, they are also linearly independent. All rows which are different from zero ( ( y i , ∙ ) {\displaystyle (y_{i},\bullet )} and ( 0 , z i ) {\displaystyle (0,z_{i})} ) are a basis of H, so there are dim ⁡ ( U ∩ W ) {\displaystyle \dim(U\cap W)} such z i {\displaystyle z_{i}} s. Therefore, the z i {\displaystyle z_{i}} s form a basis of U ∩ W {\displaystyle U\cap W} . == Example == Consider the two subspaces U = ⟨ ( 1 − 1 0 1 ) , ( 0 0 1 − 1 ) ⟩ {\displaystyle U=\left\langle \left({\begin{array}{r}1\\-1\\0\\1\end{array}}\right),\left({\begin{array}{r}0\\0\\1\\-1\end{array}}\right)\right\rangle } and W = ⟨ ( 5 0 − 3 3 ) , ( 0 5 − 3 − 2 ) ⟩ {\displaystyle W=\left\langle \left({\begin{array}{r}5\\0\\-3\\3\end{array}}\right),\left({\begin{array}{r}0\\5\\-3\\-2\end{array}}\right)\right\rangle } of the vector space R 4 {\displaystyle \mathbb {R} ^{4}} . Using the standard basis, we create the following matrix of dimension ( 2 + 2 ) × ( 2 ⋅ 4 ) {\displaystyle (2+2)\times (2\cdot 4)} : ( 1 − 1 0 1 1 − 1 0 1 0 0 1 − 1 0 0 1 − 1 5 0 − 3 3 0 0 0 0 0 5 − 3 − 2 0 0 0 0 ) . {\displaystyle \left({\begin{array}{rrrrrrrr}1&-1&0&1&&1&-1&0&1\\0&0&1&-1&&0&0&1&-1\\\\5&0&-3&3&&0&0&0&0\\0&5&-3&-2&&0&0&0&0\end{array}}\right).} Using elementary row operations, we transform this matrix into the following matrix: ( 1 0 0 0 ∙ ∙ ∙ ∙ 0 1 0 − 1 ∙ ∙ ∙ ∙ 0 0 1 − 1 ∙ ∙ ∙ ∙ 0 0 0 0 1 − 1 0 1 ) {\displaystyle \left({\begin{array}{rrrrrrrrr}1&0&0&0&&\bullet &\bullet &\bullet &\bullet \\0&1&0&-1&&\bullet &\bullet &\bullet &\bullet \\0&0&1&-1&&\bullet &\bullet &\bullet &\bullet \\\\0&0&0&0&&1&-1&0&1\end{array}}\right)} (Some entries have been replaced by " ∙ {\displaystyle \bullet } " because they are irrelevant to the result.) Therefore ( ( 1 0 0 0 ) , ( 0 1 0 − 1 ) , ( 0 0 1 − 1 ) ) {\displaystyle \left(\left({\begin{array}{r}1\\0\\0\\0\end{array}}\right),\left({\begin{array}{r}0\\1\\0\\-1\end{array}}\right),\left({\begin{array}{r}0\\0\\1\\-1\end{array}}\right)\right)} is a basis of U + W {\displaystyle U+W} , and ( ( 1 − 1 0 1 ) ) {\displaystyle \left(\left({\begin{array}{r}1\\-1\\0\\1\end{array}}\right)\right)} is a basis of U ∩ W {\displaystyle U\cap W} .
Read more →