In computing, a human-readable medium or human-readable format is any encoding of data or information that can be naturally read by humans, resulting in human-readable data. It is often encoded as ASCII or Unicode text, rather than as binary data. In most contexts, the alternative to a human-readable representation is a machine-readable format or medium of data primarily designed for reading by electronic, mechanical or optical devices, or computers. For example, Universal Product Code (UPC) barcodes are very difficult to read for humans, but very effective and reliable with the proper equipment, whereas the strings of numerals that commonly accompany the label are the human-readable form of the barcode information. Since any type of data encoding can be parsed by a suitably programmed computer, the decision to use binary encoding rather than text encoding is usually made to conserve storage space. Encoding data in a binary format typically requires fewer bytes of storage and increases efficiency of access (input and output) by eliminating format parsing or conversion. With the advent of standardized, highly structured markup languages, such as Extensible Markup Language (XML), the decreasing costs of data storage, and faster and cheaper data communication networks, compromises between human-readability and machine-readability are now more common-place than they were in the past. This has led to humane markup languages and modern configuration file formats that are far easier for humans to read. In addition, these structured representations can be compressed very effectively for transmission or storage. Human-readable protocols greatly reduce the cost of debugging. Various organizations have standardized the definition of human-readable and machine-readable data and how they are applied in their respective fields of application, e.g., the Universal Postal Union. Often the term human-readable is also used to describe shorter names or strings, that are easier to comprehend or to remember than long, complex syntax notations, such as some Uniform Resource Locator strings. Occasionally "human-readable" is used to describe ways of encoding an arbitrary integer into a long series of English words. Compared to decimal or other compact binary-to-text encoding systems, English words are easier for humans to read, remember, and type in.
Confidential computing
Confidential computing is a security and privacy-enhancing computational technique focused on protecting data in use. Confidential computing can be used in conjunction with storage and network encryption, which protect data at rest and data in transit respectively. It is designed to address software, protocol, cryptographic, and basic physical and supply-chain attacks, although some critics have demonstrated architectural and side-channel attacks effective against the technology. The technology protects data in use by performing computations in a hardware-based trusted execution environment (TEE). Confidential data is released to the TEE only once it is assessed to be trustworthy. Different types of confidential computing define the level of data isolation used, whether virtual machine, application, or function, and the technology can be deployed in on-premise data centers, edge locations, or the public cloud. It is often compared with other privacy-enhancing computational techniques such as fully homomorphic encryption, secure multi-party computation, and Trusted Computing. Confidential computing is promoted by the Confidential Computing Consortium (CCC) industry group, whose membership includes major providers of the technology. == Properties == Trusted execution environments (TEEs) "prevent unauthorized access or modification of applications and data while they are in use, thereby increasing the security level of organizations that manage sensitive and regulated data". Trusted execution environments can be instantiated on a computer's processing components such as a central processing unit (CPU) or a graphics processing unit (GPU). In their various implementations, TEEs can provide different levels of isolation including virtual machine, individual application, or compute functions. Typically, data in use in a computer's compute components and memory exists in a decrypted state and can be vulnerable to examination or tampering by unauthorized software or administrators. According to the CCC, confidential computing protects data in use through a minimum of three properties: Data confidentiality: "Unauthorized entities cannot view data while it is in use within the TEE". Data integrity: "Unauthorized entities cannot add, remove, or alter data while it is in use within the TEE". Code integrity: "Unauthorized entities cannot add, remove, or alter code executing in the TEE". In addition to trusted execution environments, remote cryptographic attestation is an essential part of confidential computing. The attestation process assesses the trustworthiness of a system and helps ensure that confidential data is released to a TEE only after it presents verifiable evidence that it is genuine and operating with an acceptable security posture. It allows the verifying party to assess the trustworthiness of a confidential computing environment through an "authentic, accurate, and timely report about the software and data state" of that environment. "Hardware-based attestation schemes rely on a trusted hardware component and associated firmware to execute attestation routines in a secure environment". Without attestation, a compromised system could deceive others into trusting it, claim it is running certain software in a TEE, and potentially compromise the confidentiality or integrity of the data being processed or the integrity of the trusted code. == Technical approaches == Technical approaches to confidential computing may vary in which software, infrastructure and administrator elements are allowed to access confidential data. The "trust boundary," which circumscribes a trusted computing base (TCB), defines which elements have the potential to access confidential data, whether they are acting benignly or maliciously. Confidential computing implementations enforce the defined trust boundary at a specific level of data isolation. The three main types of confidential computing are: Virtual machine isolation Application isolation, also known as process isolation Function isolation, also known as library isolation Virtual machine isolation removes the elements controlled by the computer infrastructure or cloud provider, but allows potential data access by elements inside a virtual machine running on the infrastructure. Application or process isolation permits data access only by authorized software applications or processes. Function or library isolation is designed to permit data access only by authorized subroutines or modules within a larger application, blocking access by any other system element, including unauthorized code in the larger application. == Threat model == As confidential computing is concerned with the protection of data in use, only certain threat models can be addressed by this technique. Other types of attacks are better addressed by other privacy-enhancing technologies. === In scope === The following threat vectors are generally considered in scope for confidential computing: Software attacks: including attacks on the host’s software and firmware. This may include the operating system, hypervisor, BIOS, other software and workloads. Protocol attacks: including "attacks on protocols associated with attestation as well as workload and data transport". This includes vulnerabilities in the "provisioning or placement of the workload" or data that could cause a compromise. Cryptographic attacks: including "vulnerabilities found in ciphers and algorithms due to a number of factors, including mathematical breakthroughs, availability of computing power and new computing approaches such as quantum computing". The CCC notes several caveats in this threat vector, including relative difficulty of upgrading cryptographic algorithms in hardware and recommendations that software and firmware be kept up-to-date. A multi-faceted, defense-in-depth strategy is recommended as a best practice. Basic physical attacks: including cold boot attacks, bus and cache snooping and plugging attack devices into an existing port, such as a PCI Express slot or USB port. Basic upstream supply-chain attacks: including attacks that would compromise TEEs through changes such as added debugging ports. The degree and mechanism of protection against these threats varies with specific confidential computing implementations. === Out of scope === Threats generally defined as out of scope for confidential computing include: Sophisticated physical attacks: including physical attacks that "require long-term and/or invasive access to hardware" such as chip scraping techniques and electron microscope probes. Upstream hardware supply-chain attacks: including attacks on the CPU manufacturing process, CPU supply chain in key injection/generation during manufacture. Attacks on components of a host system that are not directly providing the capabilities of the trusted execution environment are also generally out-of-scope. Availability attacks: confidential computing is designed to protect the confidentiality and integrity of protected data and code. It does not address availability attacks such as Denial of Service or Distributed Denial of Service attacks. == Use cases == Confidential computing can be deployed in the public cloud, on-premise data centers, or distributed "edge" locations, including network nodes, branch offices, industrial systems and others. === Data privacy and security === Confidential computing protects the confidentiality and integrity of data and code from the infrastructure provider, unauthorized or malicious software and system administrators, and other cloud tenants, which may be a concern for organizations seeking control over sensitive or regulated data. The additional security capabilities offered by confidential computing can help accelerate the transition of more sensitive workloads to the cloud or edge locations. === Multi-party analytics === Confidential computing can enable multiple parties to engage in joint analysis using confidential or regulated data inside a TEE while preserving privacy and regulatory compliance. In this case, all parties benefit from the shared analysis, but no party's sensitive data or confidential code is exposed to the other parties or system host. Examples include multiple healthcare organizations contributing data to medical research, or multiple banks collaborating to identify financial fraud or money laundering. Oxford University researchers proposed the alternative paradigm called "Confidential Remote Computing" (CRC), which supports confidential operations in Trusted Execution Environments across endpoint computers considering multiple stakeholders as mutually distrustful data, algorithm and hardware providers. === Confidential generative AI === Confidential computing technologies can be applied to various stages of a generative AI deployments to help increase data or model privacy, security, and regulatory compliance. TEEs and remote attestation can protect the integrity of data during AI model training, keep
Rhetorical structure theory
Rhetorical structure theory (RST) is a theory of text organization that describes relations that hold between parts of text. It was originally developed by William Mann, Sandra Thompson, Christian M. I. M. Matthiessen and others at the University of Southern California's Information Sciences Institute (ISI) and defined in a 1988 paper. The theory was developed as part of studies of computer-based text generation. Natural language processing researchers later began using RST in automatic summarization and other applications. It explains coherence by postulating a hierarchical, connected structure of texts, which are labeled using a small, predefined inventory of relation types - for example, one part of a text may provide an elaboration on another part, provide background or specify a cause for another. In the 2000s, following the release of the first large-scale dataset implementing the theory, the RST Discourse Treebank (RST-DT), Daniel Marcu demonstrated the feasibility of practical applications of RST to discourse parsing and summarization at ISI. Originally limited to written text, subsequent work in the 2010s expanded RST to spoken language analysis, and the framework has been applied to a variety of languages including Farsi, German, Mandarin Chinese, Russian and Spanish. Following the introduction of Transformers, LLMs have been applied to automatic RST parsing, with results approaching human performance on parsing text in English. == Rhetorical relations == Rhetorical relations, also called coherence or discourse relations, are paratactic (coordinate) or hypotactic (subordinate) relations that hold across two or more text spans. The logical arrangement of relations in a text contributes to its coherence by connecting different propositions in a relational structure. RST using rhetorical relations provides a systematic way for an analyst to analyze the underlying intention of a text. The analysis is usually built by reading the text and constructing a tree using the relations. The following example is a title and summary, appearing at the top of an article in Scientific American magazine (adapted from Ramachandran and Anstis, 1986). The original text, broken into numbered units, is: [Title:] The Perception of Apparent Motion [Abstract:] When the motion of an intermittently seen object is ambiguous the visual system resolves confusion by applying some tricks that reflect a built-in knowledge of properties of the physical world. In the figure, the numbers 1-5 show the corresponding units from the text above. Unit 5 provides an "elaboration" on unit 4, and therefore constitutes a less prominent satellite of unit 4, which acts as a nucleus for the relation. Units 4-5 form a relation "Means", explaining the means by which the visual system resolves confusion. Unit 3 is the Central Discourse Unit (CDU) of the text, since all units point to it directly or indirectly. Similarly units 1 and 2 form "preparation" and "circumstance" relations relative to their nuclei. Groups of units which serve as a satellite or nucleus together are called complex discourse units, and always span a set of adjacent EDUs. == Nuclearity in discourse == RST establishes two different types of units. Nuclei are considered as the most important parts of text whereas satellites contribute to the nuclei and are secondary. Nucleus contains basic information and satellite contains additional information about nucleus. The satellite is often incomprehensible without nucleus, whereas a text where satellites have been deleted can be understood to a certain extent. == Hierarchy in the analysis == RST relations are applied recursively in a text, until all units in that text are constituents in an RST relation. The result of such analyses is that RST structure are typically represented as trees, with one top level relation that encompasses other relations at lower levels. == Why RST? == From linguistic point of view, RST proposes a different view of text organization than most linguistic theories. RST points to a tight relation between relations and coherence in text From a computational point of view, it provides a characterization of text relations that has been implemented in different systems and for applications as text generation and summarization. == In design rationale == Computer scientists Ana Cristina Bicharra Garcia and Clarisse Sieckenius de Souz have used RST as the basis of a design rationale system called ADD+. In ADD+, RST is used as the basis for the rhetorical organization of a knowledge base, in a way comparable to other knowledge representation systems such as issue-based information system (IBIS). Similarly, RST has been used in representation schemes for argumentation.
Euratlas
Euratlas is a Switzerland-based software company dedicated to elaborate digital history maps of Europe. Founded in 2001, Euratlas has created a collection of history maps of Europe from year 1 AD to year 2000 AD that present the evolution of every country from the Roman Empire to present times. The evolution includes sovereign states and their administrative subdivisions, but also unorganized peoples and dependent territories. The maps show European country borders at regular intervals of 100 years, but not year by year. This leaves out many important turning points in history. Euratlas is considered a digital humanities company, and a scholar research software used in the field of historic cartography. It is broadly known among American and European universities, who mainly use Euratlas as a research tool and as a digital library atlas. == Sequential mapping policy == This concept was first designed by the German scholar Christian Kruse (1753–1827). Kruse, well aware that historical accounts are often biased for geographical, philosophical or political reasons, created a set of sequential maps in order to give a global vision of the successive political situations. Nowadays, the majority of atlases don't use this approach, but are event-based, like the well-known Penguin Atlas of History. The sequential approach intends to make the sequence of maps more neutral and suitable for students, historians and professionals of several fields. Although, this approach has been discussed as it leaves out many important history events that are not reflected on any of the maps because of the century interval. == Geo-referenced historical data == Initially, the European maps by century were developed as vector maps. From 2006 on, they have been converted to a geographic information system (GIS) database, enabling geo-referenced data capabilities. The map information is distributed in several layers: physical (geography information layer); political information layer (supranational entities, sovereign states, administrative divisions, dependent states and autonomous peoples); and special layers for cities and uncertain borders. The software database also contains much non-geographical information about political relationships between the various kinds of territories. == Map projection == Euratlas History Maps uses a Mercator projection, with the center in Europe. The maps include the North-African coast and the Near-East, offering a complete view of the Mediterranean Basin. The European Russia plains are shown, but not Scandinavia, specially Finland, which is cropped off the map view.
Natural language understanding
Natural language understanding (NLU) or natural language interpretation (NLI) is a subset of natural language processing in artificial intelligence that deals with machine reading comprehension. NLU has been considered an AI-hard problem. There is considerable commercial interest in the field because of its application to automated reasoning, machine translation, question answering, news-gathering, text categorization, voice-activation, archiving, and large-scale content analysis. == History == The program STUDENT, written in 1964 by Daniel Bobrow for his PhD dissertation at MIT, is one of the earliest known attempts at NLU by a computer. Eight years after John McCarthy coined the term artificial intelligence, Bobrow's dissertation (titled Natural Language Input for a Computer Problem Solving System) showed how a computer could understand simple natural language input to solve algebra word problems. A year later, in 1965, Joseph Weizenbaum at MIT wrote ELIZA, an interactive program that carried on a dialogue in English on any topic, the most popular being psychotherapy. ELIZA worked by simple parsing and substitution of key words into canned phrases and Weizenbaum sidestepped the problem of giving the program a database of real-world knowledge or a rich lexicon. Yet ELIZA gained surprising popularity as a toy project and can be seen as a very early precursor to current commercial systems such as those used by Ask.com. In 1969, Roger Schank at Stanford University introduced the conceptual dependency theory for NLU. This model, partially influenced by the work of Sydney Lamb, was extensively used by Schank's students at Yale University, such as Robert Wilensky, Wendy Lehnert, and Janet Kolodner. In 1970, William A. Woods introduced the augmented transition network (ATN) to represent natural language input. Instead of phrase structure rules ATNs used an equivalent set of finite-state automata that were called recursively. ATNs and their more general format called "generalized ATNs" continued to be used for a number of years. In 1971, Terry Winograd finished writing SHRDLU for his PhD thesis at MIT. SHRDLU could understand simple English sentences in a restricted world of children's blocks to direct a robotic arm to move items. The successful demonstration of SHRDLU provided significant momentum for continued research in the field. Winograd continued to be a major influence in the field with the publication of his book Language as a Cognitive Process. At Stanford, Winograd would later advise Larry Page, who co-founded Google. In the 1970s and 1980s, the natural language processing group at SRI International continued research and development in the field. A number of commercial efforts based on the research were undertaken, e.g., in 1982 Gary Hendrix formed Symantec Corporation originally as a company for developing a natural language interface for database queries on personal computers. However, with the advent of mouse-driven graphical user interfaces, Symantec changed direction. A number of other commercial efforts were started around the same time, e.g., Larry R. Harris at the Artificial Intelligence Corporation and Roger Schank and his students at Cognitive Systems Corp. In 1983, Michael Dyer developed the BORIS system at Yale which bore similarities to the work of Roger Schank and W. G. Lehnert. The third millennium saw the introduction of systems using machine learning for text classification, such as the IBM Watson. However, experts debate how much "understanding" such systems demonstrate: e.g., according to John Searle, Watson did not even understand the questions. John Ball, cognitive scientist and inventor of the Patom Theory, supports this assessment. Natural language processing has made inroads for applications to support human productivity in service and e-commerce, but this has largely been made possible by narrowing the scope of the application. There are thousands of ways to request something in a human language that still defies conventional natural language processing. According to Wibe Wagemans, "To have a meaningful conversation with machines is only possible when we match every word to the correct meaning based on the meanings of the other words in the sentence – just like a 3-year-old does without guesswork." == Scope and context == The umbrella term "natural language understanding" can be applied to a diverse set of computer applications, ranging from small, relatively simple tasks such as short commands issued to robots, to highly complex endeavors such as the full comprehension of newspaper articles or poetry passages. Many real-world applications fall between the two extremes, for instance text classification for the automatic analysis of emails and their routing to a suitable department in a corporation does not require an in-depth understanding of the text, but needs to deal with a much larger vocabulary and more diverse syntax than the management of simple queries to database tables with fixed schemata. Throughout the years various attempts at processing natural language or English-like sentences presented to computers have taken place at varying degrees of complexity. Some attempts have not resulted in systems with deep understanding, but have helped overall system usability. For example, Wayne Ratliff originally developed the Vulcan program with an English-like syntax to mimic the English speaking computer in Star Trek. Vulcan later became the dBase system whose easy-to-use syntax effectively launched the personal computer database industry. Systems with an easy-to-use or English-like syntax are, however, quite distinct from systems that use a rich lexicon and include an internal representation (often as first order logic) of the semantics of natural language sentences. Hence the breadth and depth of "understanding" aimed at by a system determine both the complexity of the system (and the implied challenges) and the types of applications it can deal with. The "breadth" of a system is measured by the sizes of its vocabulary and grammar. The "depth" is measured by the degree to which its understanding approximates that of a fluent native speaker. At the narrowest and shallowest, English-like command interpreters require minimal complexity, but have a small range of applications. Narrow but deep systems explore and model mechanisms of understanding, but they still have limited application. Systems that attempt to understand the contents of a document such as a news release beyond simple keyword matching and to judge its suitability for a user are broader and require significant complexity, but they are still somewhat shallow. Systems that are both very broad and very deep are beyond the current state of the art. == Components and architecture == Regardless of the approach used, most NLU systems share some common components. The system needs a lexicon of the language and a parser and grammar rules to break sentences into an internal representation. The construction of a rich lexicon with a suitable ontology requires significant effort, e.g., the Wordnet lexicon required many person-years of effort. The system also needs theory from semantics to guide the comprehension. The interpretation capabilities of a language-understanding system depend on the semantic theory it uses. Competing semantic theories of language have specific trade-offs in their suitability as the basis of computer-automated semantic interpretation. These range from naive semantics or stochastic semantic analysis to the use of pragmatics to derive meaning from context. Semantic parsers convert natural-language texts into formal meaning representations. Advanced applications of NLU also attempt to incorporate logical inference within their framework. This is generally achieved by mapping the derived meaning into a set of assertions in predicate logic, then using logical deduction to arrive at conclusions. Therefore, systems based on functional languages such as Lisp need to include a subsystem to represent logical assertions, while logic-oriented systems such as those using the language Prolog generally rely on an extension of the built-in logical representation framework. The management of context in NLU can present special challenges. A large variety of examples and counter examples have resulted in multiple approaches to the formal modeling of context, each with specific strengths and weaknesses.
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
Semantic interpretation
Semantic interpretation is an important component in dialog systems. It is related to natural language understanding, but mostly it refers to the last stage of understanding. The goal of interpretation is binding the user utterance to concept, or something the system can understand. Typically it is creating a database query based on user utterance.