A library classification is a system used within a library to organize materials, including books, sound and video recordings, electronic materials, etc., both on shelves and in catalogs and indexes. Each item is typically assigned a call number, which identifies the location of the item within the system. Materials can be arranged by many different factors, typically in either a hierarchical tree structure based on the subject or using a faceted classification system, which allows the assignment of multiple classifications to an object, enabling the classifications to be ordered in many ways. == Description == Library classification is an important and crucial aspect in library and information science. It is distinct from scientific classification in that it has as its goal to provide a useful ordering of documents rather than a theoretical organization of knowledge. Although it has the practical purpose of creating a physical ordering of documents, it does generally attempt to adhere to accepted scientific knowledge. Library classification helps to accommodate all the newly published literature in an already created order of arrangement in a filial sequence. Library classification can be defined as the arrangement of books on shelves, or description of them, in the manner which is most useful to those who read with the ultimate aim of grouping similar things together. Library classification is meant to achieve these four purposes: ordering the fields of knowledge in a systematic way, bring related items together in the most helpful sequence, provide orderly access on the shelf, and provide a location for an item on the shelf. Library classification is distinct from the application of subject headings in that classification organizes knowledge into a systematic order, while subject headings provide access to intellectual materials through vocabulary terms that may or may not be organized as a knowledge system. The characteristics that a bibliographic classification demands for the sake of reaching these purposes are: a useful sequence of subjects at all levels, a concise memorable notation, and a host of techniques and devices of number synthesis. == History == Library classifications were preceded by classifications used by bibliographers such as Conrad Gessner. The earliest library classification schemes organized books in broad subject categories. The earliest known library classification scheme is the Pinakes by Callimachus, a scholar at the Library of Alexandria during the third century BC. During the Renaissance and Reformation era, "Libraries were organized according to the whims or knowledge of individuals in charge." This changed the format in which various materials were classified. Some collections were classified by language and others by how they were printed. After the printing revolution in the sixteenth century, the increase in available printed materials made such broad classification unworkable, and more granular classifications for library materials had to be developed in the nineteenth century. In 1627 Gabriel Naudé published a book called Advice on Establishing a Library. At the time, he was working in the private library of Président à mortier Henri de Mesmes II. Mesmes had around 8,000 printed books and many more Greek, Latin and French written manuscripts. Although it was a private library, scholars with references could access it. The purpose of Advice on Establishing a Library was to identify rules for private book collectors to organize their collections in a more orderly way to increase the collection's usefulness and beauty. Naudé developed a classification system based on seven different classes: theology, medicine, jurisprudence, history, philosophy, mathematics, and the humanities. These seven classes would later be increased to twelve. Advice on Establishing a Library was about a private library, but within the same book, Naudé encouraged the idea of public libraries open to all people regardless of their ability to pay for access to the collection. One of the most famous libraries that Naudé helped improve was the Bibliothèque Mazarine in Paris. Naudé spent ten years there as a librarian. Because of Naudé's strong belief in free access to libraries to all people, the Bibliothèque Mazarine became the first public library in France around 1644. Although libraries created order within their collections from as early as the fifth century BC, the Paris Bookseller's classification, developed in 1842 by Jacques Charles Brunet, is generally seen as the first of the modern book classifications. Brunet provided five major classes: theology, jurisprudence, sciences and arts, belles-lettres, and history. Classification can now be seen as a provider of subject access to information in a networked environment. == Types == There are many standard systems of library classification in use, and many more have been proposed over the years. However, in general, classification systems can be divided into three types depending on how they are used: === Universal schemes === Covers all subjects, e.g. the Dewey Decimal Classification (DDC), Universal Decimal Classification (UDC), and Colon Classification (CC). === Specific classification schemes === Covers particular subjects or types of materials, e.g. Iconclass (art), British Catalogue of Music Classification, and Dickinson classification (music), or the NLM Classification (medicine). === National schemes === Specially created for certain countries, e.g. Swedish library classification system, SAB (Sveriges Allmänna Biblioteksförening). The Library of Congress Classification was designed around the collection of the US Library of Congress and has an American, European, and Christian bias. Nevertheless, it is used widely in large academic and research libraries. In terms of functionality, classification systems are often described as: === Enumerative === Subject headings are listed alphabetically, with numbers assigned to each heading in alphabetical order. === Hierarchical === Subjects are divided hierarchically, from most general to most specific. === Faceted/analytico-synthetic === Subjects are divided into mutually exclusive orthogonal facets. There are few completely enumerative systems or faceted systems; most systems are a blend but favouring one type or the other. The most common classification systems, LCC and DDC, are essentially enumerative, though with some hierarchical and faceted elements (more so for DDC), especially at the broadest and most general level. The first true faceted system was the colon classification of S. R. Ranganathan. == Methods or systems == Classification types denote the classification or categorization according to the form or characteristics or qualities of a classification scheme or schemes. Method and system has similar meaning. Method or methods or system means the classification schemes like Dewey Decimal Classification or Universal Decimal Classification. The types of classification is for identifying and understanding or education or research purposes while classification method means those classification schemes like DDC, UDC. === English language universal classification systems === The most common systems in English-speaking countries are: Dewey Decimal Classification (DDC) Library of Congress Classification (LCC) Universal Decimal Classification (UDC) Other systems include: Book Industry Standards and Communications (BISAC), originally developed for use by U.S. booksellers, has become increasingly popular in libraries. Bliss bibliographic classification used in some British libraries Colon classification (CC) Garside classification used in most libraries of University College London Gladstone Library Classification, devised by W.E. Gladstone and used exclusively at Gladstone's Library Harvard-Yenching Classification, an English classification system for Chinese language materials === Non-English universal classification systems === German Regensburger Verbundklassifikation (RVK) A system of book classification for Chinese libraries (Liu's Classification) library classification for user New Classification Scheme for Chinese Libraries Nippon Decimal Classification (NDC) Chinese Library Classification (CLC) Korean Decimal Classification (KDC) Russian Library-Bibliographical Classification (BBK) Swedish library classification system (SAB) === Universal classification systems that rely on synthesis (faceted systems) === Bliss bibliographic classification Colon classification Cutter Expansive Classification Universal Decimal Classification Newer classification systems tend to use the principle of synthesis (combining codes from different lists to represent the different attributes of a work) heavily, which is comparatively lacking in LC or DDC. == Practice == Library classification is associated with library (descriptive) cataloging under the rubric of cataloging and classification, sometimes grouped together as technical serv
Cleverbot
Cleverbot is a chatterbot web application. It was created by British AI scientist Rollo Carpenter and launched in October 2008. It was preceded by Jabberwacky, a chatbot project that began in 1988 and went online in 1997. In its first decade, Cleverbot held several thousand conversations with Carpenter and his associates. Since launching on the web, the number of conversations held has exceeded 150 million. Besides the web application, Cleverbot is also available as an iOS, Android, and Windows Phone app. == Operation == Cleverbot's responses are not pre-programmed because it learns from human input: Humans type into the box below the Cleverbot logo and the system finds all keywords or an exact phrase matching the input. After searching through its saved conversations, it responds to the input by finding how a human responded to that input when it was asked, in part or in full, by Cleverbot. Cleverbot participated in a formal Turing test at the 2011 Techniche festival at the Indian Institute of Technology Guwahati on 3 September 2011. Out of the 1334 votes cast, Cleverbot was judged to be 59.3% human, compared to the rating of 63.3% human achieved by human participants. A score of 50.05% or higher is often considered to be a passing grade. The software running for the event had to handle just 1 or 2 simultaneous requests, whereas online Cleverbot is usually talking to around 10,000 to 50,000 people at once. == Developments == Cleverbot is constantly growing in data size at the rate of 4 to 7 million interactions per day. Updates to the software have been mostly behind the scenes. In 2014, Cleverbot was upgraded to use GPU serving techniques. Unlike Eliza, the program does not respond in a fixed way, instead choosing its responses heuristically using fuzzy logic, the whole of the conversation being compared to the millions that have taken place before. Cleverbot now uses over 279 million interactions, about 3-4% of the data it has already accumulated. The developers of Cleverbot are attempting to build a new version using machine learning techniques. An app that uses the Cleverscript engine to play a game of 20 Questions has been launched under the name Clevernator. Unlike other such games, the player asks the questions and it is the role of the AI to understand, and answer factually. An app that allows owners to create and talk to their own small Cleverbot-like AI has been launched, called Cleverme! for Apple products. == In popular culture == Cleverbot received media attention after being featured in the popular 2010 creepypasta ARG web serial Ben Drowned by Alexander D. Hall. In early 2017, a Twitch stream of two Google Home devices modified to talk to each other using Cleverbot garnered over 700,000 visitors and over 30,000 peak concurrent viewers.
Hallucination (artificial intelligence)
In the field of artificial intelligence (AI), a hallucination or artificial hallucination (also called bullshitting, confabulation, or delusion) is a response generated by AI that contains false or misleading information presented as fact. This term draws a loose analogy with human psychology, where a hallucination typically involves false percepts. For example, a chatbot powered by large language models (LLMs), like ChatGPT, may embed plausible-sounding random falsehoods within its generated content. Detecting and mitigating errors and hallucinations pose significant challenges for practical deployment and reliability of LLMs in high-stakes scenarios, such as chip design, supply chain logistics, and medical diagnostics. Some software engineers and statisticians have criticized the specific term "AI hallucination" for unreasonably anthropomorphizing computers. Symbolic artificial intelligence models generally do not produce hallucinations, unlike large language models. == Term == === Origin === Since the 1980s, the term "hallucination" has been used in computer vision with a positive connotation to describe the process of adding detail to an image. For example, the task of generating high-resolution face images from low-resolution inputs is called face hallucination. The first documented use of the term "hallucination" in this sense is in the PhD thesis of Eric Mjolsness in 1986. A notable work is the face hallucination algorithm by Simon Baker and Takeo Kanade published in 1999. In the 2000s, hallucinations were described in statistical machine translation as a failure mode. Since the 2010s, the term has undergone a semantic shift to signify the generation of factually incorrect or misleading outputs by AI systems in tasks like machine translation and object detection. In 2015, hallucinations were identified in visual semantic role labeling tasks by Saurabh Gupta and Jitendra Malik. In 2015, computer scientist Andrej Karpathy used the term "hallucinated" in a blog post to describe his recurrent neural network (RNN) language model generating an incorrect citation link. In 2017, Google researchers used the term to describe the responses generated by neural machine translation (NMT) models when they are not related to the source text, and in 2018, the term was used in computer vision to describe instances where non-existent objects are erroneously detected because of adversarial attacks. In July 2021, Meta warned during its release of BlenderBot 2 that the system is prone to "hallucinations", which Meta defined as "confident statements that are not true". Following OpenAI's ChatGPT release in beta version in November 2022, some users complained that such chatbots often seem to pointlessly embed plausible-sounding random falsehoods within their generated content. Many news outlets, including The New York Times, started to use the term "hallucinations" to describe these models' frequently incorrect or inconsistent responses. In 2023, the Cambridge dictionary updated its definition of hallucination to include this new sense specific to the field of AI. Some researchers have highlighted a lack of consistency in how the term is used, but also identified several alternative terms in the literature, such as confabulations, fabrications, and factual errors. === Definitions and alternatives === Uses, definitions and characterizations of the term "hallucination" in the context of LLMs include: "a tendency to invent facts in moments of uncertainty" (OpenAI, May 2023) "a model's logical mistakes" (OpenAI, May 2023) "fabricating information entirely, but behaving as if spouting facts" (CNBC, May 2023) "making up information" (The Verge, February 2023) "probability distributions" (in scientific contexts) Journalist Benj Edwards, in Ars Technica, writes that the term "hallucination" is controversial, but that some form of metaphor remains necessary; Edwards suggests "confabulation" as an analogy for processes that involve "creative gap-filling". In July 2024, a White House report on fostering public trust in AI research mentioned hallucinations only in the context of reducing them. Notably, when acknowledging David Baker's Nobel Prize-winning work with AI-generated proteins, the Nobel committee avoided the term entirely, instead referring to "imaginative protein creation". Hicks, Humphries, and Slater, in their article in Ethics and Information Technology, argue that the output of LLMs is "bullshit" under Harry Frankfurt's definition of the term, and that the models are "in an important way indifferent to the truth of their outputs", with true statements only accidentally true, and false ones accidentally false. Some researchers also use the derogatory term "botshit", often referring to uncritical use of AI. === Criticism === In the scientific community, some researchers avoid the term "hallucination", seeing it as potentially misleading. It has been criticized by Usama Fayyad, executive director of the Institute for Experimental Artificial Intelligence at Northeastern University, on the grounds that it misleadingly personifies large language models and is vague. Mary Shaw said, "The current fashion for calling generative AI's errors 'hallucinations' is appalling. It anthropomorphizes the software, and it spins actual errors as somehow being idiosyncratic quirks of the system even when they're objectively incorrect." In Salon, statistician Gary Smith argues that LLMs "do not understand what words mean" and consequently that the term "hallucination" unreasonably anthropomorphizes the machine. Murray Shanahan argues that anthropomorphic framing of LLM capabilities, including terms like "hallucination", encourages users and researchers to attribute cognitive processes to systems that operate through statistical pattern completion, and advocates for more careful linguistic practices when discussing LLM behavior. Kristina Šekrst argues that applying psychological vocabulary to LLM outputs obscures the difference between the appearance of mental properties and their genuine presence. Förster & Skop assert that tech companies use the hallucination metaphor to anthropomorphize models and deflect responsibility for non-factual outputs. Some see the AI outputs not as illusory but as prospective—that is, having some chance of being true, similar to early-stage scientific conjectures. The term has also been criticized for its association with psychedelic drug experiences. == In natural language generation == In natural language generation, there are several reasons why natural language models hallucinate: === Hallucination from data === Hallucinations can stem from incomplete, inaccurate or unrepresentative data sets. === Modeling-related causes === The pre-training of generative pretrained transformers (GPT) involves predicting the next word. It incentivizes GPT models to "give a guess" about what the next word is, even when they lack information. Some researchers take an anthropomorphic perspective and posit that hallucinations arise from a tension between novelty and usefulness. For instance, Amabile and Pratt define human creativity as the production of novel and useful ideas. By extension, a focus on novelty in machine creativity can lead to the production of original but inaccurate responses—that is, falsehoods—whereas a focus on usefulness may result in memorized content lacking originality. By 2022, newspapers such as The New York Times expressed concern that, as the adoption of bots based on large language models continued to grow, unwarranted user confidence in bot output could lead to problems. === Interpretability research === In 2025, interpretability research by Anthropic on the LLM Claude identified internal circuits that cause it to decline to answer questions unless it knows the answer. By default, the circuit is active and the LLM doesn't answer. When the LLM has sufficient information, these circuits are inhibited and the LLM answers the question. Hallucinations were found to occur when this inhibition happens incorrectly, such as when Claude recognizes a name but lacks sufficient information about that person, causing it to generate plausible but untrue responses. === Examples === On 15 November 2022, researchers from Meta AI published Galactica, designed to "store, combine and reason about scientific knowledge". Content generated by Galactica came with the warning: "Outputs may be unreliable! Language Models are prone to hallucinate text." In one case, when asked to draft a paper on creating avatars, Galactica cited a fictitious paper from a real author who works in the relevant area. Meta withdrew Galactica on 17 November due to offensiveness and inaccuracy. OpenAI's ChatGPT, released in beta version to the public on November 30, 2022, was based on the foundation model GPT-3.5 (a revision of GPT-3). Professor Ethan Mollick of Wharton called it an "omniscient, eager-to-please intern who sometimes lies to you". Data scientist Teresa Kuba
And–or tree
An and–or tree is a graphical representation of the reduction of problems (or goals) to conjunctions and disjunctions of subproblems (or subgoals). == Example == The and–or tree: represents the search space for solving the problem P, using the goal-reduction methods: P if Q and R P if S Q if T Q if U == Definitions == Given an initial problem P0 and set of problem solving methods of the form: P if P1 and … and Pn the associated and–or tree is a set of labelled nodes such that: The root of the tree is a node labelled by P0. For every node N labelled by a problem or sub-problem P and for every method of the form P if P1 and ... and Pn, there exists a set of children nodes N1, ..., Nn of the node N, such that each node Ni is labelled by Pi. The nodes are conjoined by an arc, to distinguish them from children of N that might be associated with other methods. A node N, labelled by a problem P, is a success node if there is a method of the form P if nothing (i.e., P is a "fact"). The node is a failure node if there is no method for solving P. If all of the children of a node N, conjoined by the same arc, are success nodes, then the node N is also a success node. Otherwise the node is a failure node. == Search strategies == An and–or tree specifies only the search space for solving a problem. Different search strategies for searching the space are possible. These include searching the tree depth-first, breadth-first, or best-first using some measure of desirability of solutions. The search strategy can be sequential, searching or generating one node at a time, or parallel, searching or generating several nodes in parallel. == Relationship with logic programming == The methods used for generating and–or trees are propositional logic programs (without variables). In the case of logic programs containing variables, the solutions of conjoint sub-problems must be compatible. Subject to this complication, sequential and parallel search strategies for and–or trees provide a computational model for executing logic programs. == Relationship with two-player games == And–or trees can also be used to represent the search spaces for two-person games. The root node of such a tree represents the problem of one of the players winning the game, starting from the initial state of the game. Given a node N, labelled by the problem P of the player winning the game from a particular state of play, there exists a single set of conjoint children nodes, corresponding to all of the opponents responding moves. For each of these children nodes, there exists a set of non-conjoint children nodes, corresponding to all of the player's defending moves. For solving game trees with proof-number search family of algorithms, game trees are to be mapped to and–or trees. MAX-nodes (i.e. maximizing player to move) are represented as OR nodes, MIN-nodes map to AND nodes. The mapping is possible, when the search is done with only a binary goal, which usually is "player to move wins the game".
Machine learning in video games
Artificial intelligence and machine learning techniques are used in video games for a wide variety of applications such as non-player character (NPC) control, procedural content generation (PCG) and deep learning-based content generation. Machine learning is a subset of artificial intelligence that uses historical data to build predictive and analytical models. This is in sharp contrast to traditional methods of artificial intelligence such as search trees and expert systems. Information on machine learning techniques in the field of games is mostly known to public through research projects as most gaming companies choose not to publish specific information about their intellectual property. The most publicly known application of machine learning in games is likely the use of deep learning agents that compete with professional human players in complex strategy games. There has been a significant application of machine learning on games such as Atari/ALE, Doom, Minecraft, StarCraft, and car racing. Other games that did not originally exists as video games, such as chess and Go have also been affected by the machine learning. == Overview of relevant machine learning techniques == === Deep learning === Deep learning is a subset of machine learning which focuses heavily on the use of artificial neural networks (ANN) that learn to solve complex tasks. Deep learning uses multiple layers of ANN and other techniques to progressively extract information from an input. Due to this complex layered approach, deep learning models often require powerful machines to train and run on. ==== Convolutional neural networks ==== Convolutional neural networks (CNN) are specialized ANNs that are often used to analyze image data. These types of networks are able to learn translation invariant patterns, which are patterns that are not dependent on location. CNNs are able to learn these patterns in a hierarchy, meaning that earlier convolutional layers will learn smaller local patterns while later layers will learn larger patterns based on the previous patterns. A CNN's ability to learn visual data has made it a commonly used tool for deep learning in games. === Recurrent neural network === Recurrent neural networks are a type of ANN that are designed to process sequences of data in order, one part at a time rather than all at once. An RNN runs over each part of a sequence, using the current part of the sequence along with memory of previous parts of the current sequence to produce an output. These types of ANN are highly effective at tasks such as speech recognition and other problems that depend heavily on temporal order. There are several types of RNNs with different internal configurations; the basic implementation suffers from a lack of long term memory due to the vanishing gradient problem, thus it is rarely used over newer implementations. ==== Long short-term memory ==== A long short-term memory (LSTM) network is a specific implementation of a RNN that is designed to deal with the vanishing gradient problem seen in simple RNNs, which would lead to them gradually "forgetting" about previous parts of an inputted sequence when calculating the output of a current part. LSTMs solve this problem with the addition of an elaborate system that uses an additional input/output to keep track of long term data. LSTMs have achieved very strong results across various fields, and were used by several monumental deep learning agents in games. === Reinforcement learning === Reinforcement learning is the process of training an agent using rewards and/or punishments. The way an agent is rewarded or punished depends heavily on the problem; such as giving an agent a positive reward for winning a game or a negative one for losing. Reinforcement learning is used heavily in the field of machine learning and can be seen in methods such as Q-learning, policy search, Deep Q-networks and others. It has seen strong performance in both the field of games and robotics. === Neuroevolution === Neuroevolution involves the use of both neural networks and evolutionary algorithms. Instead of using gradient descent like most neural networks, neuroevolution models make use of evolutionary algorithms to update neurons in the network. Researchers claim that this process is less likely to get stuck in a local minimum and is potentially faster than state of the art deep learning techniques. == Deep learning agents == Machine learning agents have been used to take the place of a human player rather than function as NPCs, which are deliberately added into video games as part of designed gameplay. Deep learning agents have achieved impressive results when used in competition with both humans and other artificial intelligence agents. === Chess === Chess is a turn-based strategy game that is considered a difficult AI problem due to the computational complexity of its board space. Similar strategy games are often solved with some form of a Minimax Tree Search. These types of AI agents have been known to beat professional human players, such as the historic 1997 Deep Blue versus Garry Kasparov match. Since then, machine learning agents have shown ever greater success than previous AI agents. === Go === Go is another turn-based strategy game which is considered an even more difficult AI problem than chess. The state space of is Go is around 10^170 possible board states compared to the 10^120 board states for Chess. Prior to recent deep learning models, AI Go agents were only able to play at the level of a human amateur. ==== AlphaGo ==== Google's 2015 AlphaGo was the first AI agent to beat a professional Go player. AlphaGo used a deep learning model to train the weights of a Monte Carlo tree search (MCTS). The deep learning model consisted of 2 ANN, a policy network to predict the probabilities of potential moves by opponents, and a value network to predict the win chance of a given state. The deep learning model allows the agent to explore potential game states more efficiently than a vanilla MCTS. The network were initially trained on games of humans players and then were further trained by games against itself. ==== AlphaGo Zero ==== AlphaGo Zero, another implementation of AlphaGo, was able to train entirely by playing against itself. It was able to quickly train up to the capabilities of the previous agent. === StarCraft series === StarCraft and its sequel StarCraft II are real-time strategy (RTS) video games that have become popular environments for AI research. Blizzard and DeepMind have worked together to release a public StarCraft 2 environment for AI research to be done on. Various deep learning methods have been tested on both games, though most agents usually have trouble outperforming the default AI with cheats enabled or skilled players of the game. ==== Alphastar ==== Alphastar was the first AI agent to beat professional StarCraft 2 players without any in-game advantages. The deep learning network of the agent initially received input from a simplified zoomed out version of the gamestate, but was later updated to play using a camera like other human players. The developers have not publicly released the code or architecture of their model, but have listed several state of the art machine learning techniques such as relational deep reinforcement learning, long short-term memory, auto-regressive policy heads, pointer networks, and centralized value baseline. Alphastar was initially trained with supervised learning, it watched replays of many human games in order to learn basic strategies. It then trained against different versions of itself and was improved through reinforcement learning. The final version was hugely successful, but only trained to play on a specific map in a protoss mirror matchup. === Dota 2 === Dota 2 is a multiplayer online battle arena (MOBA) game. Like other complex games, traditional AI agents have not been able to compete on the same level as professional human player. The only widely published information on AI agents attempted on Dota 2 is OpenAI's deep learning Five agent. ==== OpenAI Five ==== OpenAI Five utilized separate long short-term memory networks to learn each hero. It trained using a reinforcement learning technique known as Proximal Policy Learning running on a system containing 256 GPUs and 128,000 CPU cores. Five trained for months, accumulating 180 years of game experience each day, before facing off with professional players. It was eventually able to beat the 2018 Dota 2 esports champion team in a 2019 series of games. === Planetary Annihilation === Planetary Annihilation is a real-time strategy game which focuses on massive scale war. The developers use ANNs in their default AI agent. === Supreme Commander 2 === Supreme Commander 2 is a real-time strategy (RTS) video game. The game uses Multilayer Perceptrons (MLPs) to control a platoon’s reaction to encountered enemy units. Total of four MLPs are used, one for each platoon type: land, naval
Companion robot
A companion robot is a robot created to create real or apparent companionship for human beings. Target markets for companion robots include the elderly and single children. Companions robots are expected to communicate with non-experts in a natural and intuitive way. They offer a variety of functions, such as monitoring the home remotely, communicating with people, or waking people up in the morning. Their aim is to perform a wide array of tasks including educational functions, home security, diary duties, entertainment and message delivery services, etc. The idea of companionship with robots has already existed on science fictions of 1970s, like R2-D2. Starting from the late 20th century, companion robots became a reality, mostly as robotic pets. Besides entertainment purposes, interactive robots were also introduced as a personal service robot for elderly care around 2000. == Characteristics == Companion robots try to interact with users. They gather information about users based on their interactions and yield feedback. This procedure varies slightly based on their specific roles. For example, social-companion robots make simple conversations, while pet-companion robots mimic being real pets. == Types == Companion robots can perform a variety of tasks and they are produced in a specialized manner according to their purpose or target audience in order to increase convenience and end user satisfaction. === Social companion robots === Social companion robots are designed to provide companionship and be a solution for unwanted solitude. They often mimic adult human, child or pet behaviours appealing to the user base. Robots which are specifically devised for simple conversations, conveying emotions and respond to user feelings fall under this category. === Assistive companion robots === Assistive companion robots are aimed at people who require constant care because of age, disability or rehabilitation purposes. Such robots can help disadvantaged users with their daily tasks, act as reminders (e.g., for regular medication) and facilitate mobility in everyday actions. Assistive companion robots reduce the intensity of labour that should be performed by caretakers, nurses and legal guardians. === Educational companion robots === Educational companion robots perform tutorship for students, regardless of their ages, and can teach desired subjects with activities tailored for the user such as interactive assignments and games. Rather than replacing teachers and instructors, educational companion robots are aides to them. === Therapeutic companion robots === Designed for individuals coping with stress (PTSD in severe cases), anxiety and loneliness; therapeutic companion robots support users' emotional and mental wellbeing. Such robots can be utilized in hospitals and care facilities as well as dwellings where the distressed user may need the most help. Therapeutic companion robots bear a vast resemblance to assistive companion robots to the extent of being a branch of them; the nuance between these two types of companion robots is that the former is for long-term/lifetime usage while the latter is mostly for the duration of the therapy received by the user. === Pet companion robots === Pet companion robots are for individuals who seek an alternative to live pets as live animals demand a considerable amount of care and may not be eligible for people with allergies. These robots aim to be perfect imitations of a pet while diminishing the chore aspect of having one. === Entertainment companion robots === Entertainment companion robots are designed solely for entertainment and can provide numerous ways of entertainment, ranging from dancing to playing games with the user. People who would appreciate an individual to have fun with are the main audience of such products. === Personal assistant robots === Personal assistant robots help people with daily tasks, management, scheduling, reminding etc. Their area of activity can be offices as well as homes and public spaces. === Sex robots === Sex robots are anthropomorphic robotic sex dolls that have human-like movement or behavior, and some degree of artificial intelligence. As of 2026, although elaborately instrumented sex dolls have been created by a number of inventors, no fully animated sex robots yet exist. Simple devices have been created which can speak, make facial expressions, or respond to touch. There is controversy as to whether developing them would be morally justifiable. In 2015, robot ethicist Kathleen Richardson called for a ban on the creation of anthropomorphic sex robots with concerns about normalizing relationships with machines and reinforcing female dehumanization. Questions about their ethics, effects, and possible legal regulations have been discussed since then. == Examples == There are several companion robot prototypes, and these include Paro, CompanionAble, and EmotiRob, among others. === Paro === Paro is a pet-type robot system developed by Japan's National Institute of Advanced Industrial Science and Technology (AIST). The robot, which looked like a small harp seal, was designed as a therapeutic tool for use in hospitals and nursing homes. The robot is programmed to cry for attention and respond to its name. Experiments showed that Paro facilitated elderly residents to communicate with each other, which led to psychological improvements. === CompanionAble === This robot is classified as an FP 7 EU project. It is built to "cooperate with Ambient Assistive Living environment". The autonomous device, which is also built to support the elderly, helps its owner interact with smart home environment as well as caregivers. The robot functions as a mobile friend, by which natural interaction is possible via speech and the touchscreen to detect and track people at home. === EmotiRob === EmotiRob is developed in a robotics project which is the continuity of the MAPH (Active Media For the Handicap) project in emotion synthesis. The aim of the project was to maintain emotional interaction with children. EmotiRob designed in a way that a child can hold it in a his/her arms and with which he/she could interact by talking to it, and then the robot would express itself through body postures or facial expressions. It has cognitive capabilities, which are further extended so that the robot can have a natural linguistic interaction with its owner through the DRAGON speech-recognition software developed by a company called NUANCE. Such interaction is expected to facilitate a child's cognitive development and develop new learning patterns. === LOVOT === Lovot is a Japanese company robot whose only purpose is "to make you happy". It features over 50 sensors that mimic the behavior of a human baby or small pet, a 360° camera with a microphone, the ability to distinguish humans from objects, neoteny eyes, and an internal warmth of 30° celsius. An interactive Lovot Café was opened in Japan October 3, 2020. === NICOBO === Nicobo was developed by Panasonic and was influenced by the loneliness of lockdowns created as a measure of the COVID-19 pandemic. It was designed to appear vulnerable, which creates empathy in its owners. Nicobo's name derives from the Japanese word for "smile". It wags its tail, engages in baby talk, and stays as a housemate. === Hyodol === Hyodol is an advanced care robot designed to support the elderly by reminding them to take their medications and monitoring their movements to keep their guardians informed. Additionally, this innovative robot can detect and respond to the emotional states of its elderly users, adding a layer of personalized care. Hyodol is designed with the appearance and speech style of a 7-year-old Korean grandchild, featuring a soft fabric exterior and user interaction methods such as striking the head or patting the back. It is equipped with various sensors and wireless communication technologies to collect and process data, supporting mobile apps and PC web monitoring systems for remote monitoring from anywhere. In South Korea, approximately 10,000 Hyodol robots are deployed to the homes of elderly individuals living alone, providing essential support and companionship. Local governments, including provincial and county offices, have embraced Hyodol as a solution to address social challenges stemming from the country's rapidly aging society.Furthermore, the robot is widely utilized in the treatment of dementia patients at a university hospital in Gangwon province. Hyodol was honored with the Mobile World Congress (MWC) Global Mobile Awards (GLOMO) in the "Best Mobile Innovation for Connected Health and Wellbeing" category on February 29, 2024. === Moxie === Moxie was a companion robot for autistic children developed by a company called Embodied. Although it had limited motion, it presented itself as a lifelike avatar. It was designed to help the children learn emotional cognition, using remotely hosted large language models to direct its respons
Feature hashing
In machine learning, feature hashing, also known as the hashing trick (by analogy to the kernel trick), is a fast and space-efficient way of vectorizing features, i.e. turning arbitrary features into indices in a vector or matrix. It works by applying a hash function to the features and using their hash values as indices directly (after a modulo operation), rather than looking the indices up in an associative array. In addition to its use for encoding non-numeric values, feature hashing can also be used for dimensionality reduction. This trick is often attributed to Weinberger et al. (2009), but there exists a much earlier description of this method published by John Moody in 1989. == Motivation == === Motivating example === In a typical document classification task, the input to the machine learning algorithm (both during learning and classification) is free text. From this, a bag of words (BOW) representation is constructed: the individual tokens are extracted and counted, and each distinct token in the training set defines a feature (independent variable) of each of the documents in both the training and test sets. Machine learning algorithms, however, are typically defined in terms of numerical vectors. Therefore, the bags of words for a set of documents is regarded as a term-document matrix where each row is a single document, and each column is a single feature/word; the entry i, j in such a matrix captures the frequency (or weight) of the j'th term of the vocabulary in document i. (An alternative convention swaps the rows and columns of the matrix, but this difference is immaterial.) Typically, these vectors are extremely sparse—according to Zipf's law. The common approach is to construct, at learning time or prior to that, a dictionary representation of the vocabulary of the training set, and use that to map words to indices. Hash tables and tries are common candidates for dictionary implementation. E.g., the three documents John likes to watch movies. Mary likes movies too. John also likes football. can be converted, using the dictionary to the term-document matrix ( John likes to watch movies Mary too also football 1 1 1 1 1 0 0 0 0 0 1 0 0 1 1 1 0 0 1 1 0 0 0 0 0 1 1 ) {\displaystyle {\begin{pmatrix}{\textrm {John}}&{\textrm {likes}}&{\textrm {to}}&{\textrm {watch}}&{\textrm {movies}}&{\textrm {Mary}}&{\textrm {too}}&{\textrm {also}}&{\textrm {football}}\\1&1&1&1&1&0&0&0&0\\0&1&0&0&1&1&1&0&0\\1&1&0&0&0&0&0&1&1\end{pmatrix}}} (Punctuation was removed, as is usual in document classification and clustering.) The problem with this process is that such dictionaries take up a large amount of storage space and grow in size as the training set grows. On the contrary, if the vocabulary is kept fixed and not increased with a growing training set, an adversary may try to invent new words or misspellings that are not in the stored vocabulary so as to circumvent a machine learned filter. To address this challenge, Yahoo! Research attempted to use feature hashing for their spam filters. Note that the hashing trick isn't limited to text classification and similar tasks at the document level, but can be applied to any problem that involves large (perhaps unbounded) numbers of features. === Mathematical motivation === Mathematically, a token is an element t {\displaystyle t} in a finite (or countably infinite) set T {\displaystyle T} . Suppose we only need to process a finite corpus, then we can put all tokens appearing in the corpus into T {\displaystyle T} , meaning that T {\displaystyle T} is finite. However, suppose we want to process all possible words made of the English letters, then T {\displaystyle T} is countably infinite. Most neural networks can only operate on real vector inputs, so we must construct a "dictionary" function ϕ : T → R n {\displaystyle \phi :T\to \mathbb {R} ^{n}} . When T {\displaystyle T} is finite, of size | T | = m ≤ n {\displaystyle |T|=m\leq n} , then we can use one-hot encoding to map it into R n {\displaystyle \mathbb {R} ^{n}} . First, arbitrarily enumerate T = { t 1 , t 2 , . . , t m } {\displaystyle T=\{t_{1},t_{2},..,t_{m}\}} , then define ϕ ( t i ) = e i {\displaystyle \phi (t_{i})=e_{i}} . In other words, we assign a unique index i {\displaystyle i} to each token, then map the token with index i {\displaystyle i} to the unit basis vector e i {\displaystyle e_{i}} . One-hot encoding is easy to interpret, but it requires one to maintain the arbitrary enumeration of T {\displaystyle T} . Given a token t ∈ T {\displaystyle t\in T} , to compute ϕ ( t ) {\displaystyle \phi (t)} , we must find out the index i {\displaystyle i} of the token t {\displaystyle t} . Thus, to implement ϕ {\displaystyle \phi } efficiently, we need a fast-to-compute bijection h : T → { 1 , . . . , m } {\displaystyle h:T\to \{1,...,m\}} , then we have ϕ ( t ) = e h ( t ) {\displaystyle \phi (t)=e_{h(t)}} . In fact, we can relax the requirement slightly: It suffices to have a fast-to-compute injection h : T → { 1 , . . . , n } {\displaystyle h:T\to \{1,...,n\}} , then use ϕ ( t ) = e h ( t ) {\displaystyle \phi (t)=e_{h(t)}} . In practice, there is no simple way to construct an efficient injection h : T → { 1 , . . . , n } {\displaystyle h:T\to \{1,...,n\}} . However, we do not need a strict injection, but only an approximate injection. That is, when t ≠ t ′ {\displaystyle t\neq t'} , we should probably have h ( t ) ≠ h ( t ′ ) {\displaystyle h(t)\neq h(t')} , so that probably ϕ ( t ) ≠ ϕ ( t ′ ) {\displaystyle \phi (t)\neq \phi (t')} . At this point, we have just specified that h {\displaystyle h} should be a hashing function. Thus we reach the idea of feature hashing. == Algorithms == === Feature hashing (Weinberger et al. 2009) === The basic feature hashing algorithm presented in (Weinberger et al. 2009) is defined as follows. First, one specifies two hash functions: the kernel hash h : T → { 1 , 2 , . . . , n } {\displaystyle h:T\to \{1,2,...,n\}} , and the sign hash ζ : T → { − 1 , + 1 } {\displaystyle \zeta :T\to \{-1,+1\}} . Next, one defines the feature hashing function: ϕ : T → R n , ϕ ( t ) = ζ ( t ) e h ( t ) {\displaystyle \phi :T\to \mathbb {R} ^{n},\quad \phi (t)=\zeta (t)e_{h(t)}} Finally, extend this feature hashing function to strings of tokens by ϕ : T ∗ → R n , ϕ ( t 1 , . . . , t k ) = ∑ j = 1 k ϕ ( t j ) {\displaystyle \phi :T^{}\to \mathbb {R} ^{n},\quad \phi (t_{1},...,t_{k})=\sum _{j=1}^{k}\phi (t_{j})} where T ∗ {\displaystyle T^{}} is the set of all finite strings consisting of tokens in T {\displaystyle T} . Equivalently, ϕ ( t 1 , . . . , t k ) = ∑ j = 1 k ζ ( t j ) e h ( t j ) = ∑ i = 1 n ( ∑ j : h ( t j ) = i ζ ( t j ) ) e i {\displaystyle \phi (t_{1},...,t_{k})=\sum _{j=1}^{k}\zeta (t_{j})e_{h(t_{j})}=\sum _{i=1}^{n}\left(\sum _{j:h(t_{j})=i}\zeta (t_{j})\right)e_{i}} ==== Geometric properties ==== We want to say something about the geometric property of ϕ {\displaystyle \phi } , but T {\displaystyle T} , by itself, is just a set of tokens, we cannot impose a geometric structure on it except the discrete topology, which is generated by the discrete metric. To make it nicer, we lift it to T → R T {\displaystyle T\to \mathbb {R} ^{T}} , and lift ϕ {\displaystyle \phi } from ϕ : T → R n {\displaystyle \phi :T\to \mathbb {R} ^{n}} to ϕ : R T → R n {\displaystyle \phi :\mathbb {R} ^{T}\to \mathbb {R} ^{n}} by linear extension: ϕ ( ( x t ) t ∈ T ) = ∑ t ∈ T x t ζ ( t ) e h ( t ) = ∑ i = 1 n ( ∑ t : h ( t ) = i x t ζ ( t ) ) e i {\displaystyle \phi ((x_{t})_{t\in T})=\sum _{t\in T}x_{t}\zeta (t)e_{h(t)}=\sum _{i=1}^{n}\left(\sum _{t:h(t)=i}x_{t}\zeta (t)\right)e_{i}} There is an infinite sum there, which must be handled at once. There are essentially only two ways to handle infinities. One may impose a metric, then take its completion, to allow well-behaved infinite sums, or one may demand that nothing is actually infinite, only potentially so. Here, we go for the potential-infinity way, by restricting R T {\displaystyle \mathbb {R} ^{T}} to contain only vectors with finite support: ∀ ( x t ) t ∈ T ∈ R T {\displaystyle \forall (x_{t})_{t\in T}\in \mathbb {R} ^{T}} , only finitely many entries of ( x t ) t ∈ T {\displaystyle (x_{t})_{t\in T}} are nonzero. Define an inner product on R T {\displaystyle \mathbb {R} ^{T}} in the obvious way: ⟨ e t , e t ′ ⟩ = { 1 , if t = t ′ , 0 , else. ⟨ x , x ′ ⟩ = ∑ t , t ′ ∈ T x t x t ′ ⟨ e t , e t ′ ⟩ {\displaystyle \langle e_{t},e_{t'}\rangle ={\begin{cases}1,{\text{ if }}t=t',\\0,{\text{ else.}}\end{cases}}\quad \langle x,x'\rangle =\sum _{t,t'\in T}x_{t}x_{t'}\langle e_{t},e_{t'}\rangle } As a side note, if T {\displaystyle T} is infinite, then the inner product space R T {\displaystyle \mathbb {R} ^{T}} is not complete. Taking its completion would get us to a Hilbert space, which allows well-behaved infinite sums. Now we have an inner product space, with enough structure to describe the geometry of the feature hashing function ϕ : R T → R n {\displaystyle \phi :\ma