TW201804345A - Method, system and computer-readable medium for automatic chinese ontology generation based on structured web knowledge - Google Patents

Abstract

A method, system, computer device and computer-readable medium on the basis of a Chinese language ontology database automatically generated from structured network knowledge. Said method comprising the following steps: capturing structured knowledge from a structured knowledge network, wherein the structured knowledge contains at least one focus concept which is used to generate an automatic Chinese ontology database; filtering links which are not relevant; extracting knowledge related to the focus concept; discovering concepts related to said focus concept; deducing a semantic correlation between said focus concept and the concepts related thereto; and storing data of said deduced semantic correlation. The present invention provides a more efficient system and method for automatically generating a Chinese language ontology database, which answers to the world of data which is rapidly developing, and caters to the needs of data users.

Description

Method, system and computer-readable medium for constructing automatic Chinese ontology library based on structured network knowledge

The invention relates to a method and a system for automatically generating an ontology library, in particular to an automatic Chinese ontology library construction based on structured network knowledge.

In the age of information technology, large amounts of data are uploaded to or downloaded from the Internet, corporate computer networks, or other databases every day. Data users always expect to get all the information they need from the Internet, corporate computer networks, or databases, but not every time they get the right information. Ontology represents the unique similarity and connection relationship between different concepts, which can be used to help the semantic search of information or files obtained by the network, corporate computer network or any other database.

Traditional ontology generation is usually done by experts manually entering relationships between concepts, so it requires a lot of manpower. Currently, different computer-implemented programs, such as artificial neural networks (ANN), can be used to discover semantic correlations between words in a corpus. However, ANN needs to be trained in advance, so it still requires a lot of manpower to prepare data with multiple input modes. Therefore, the use of ANN may not be able to effectively keep up with the Internet, corporate computer networks or any Any database data update speed.

Ontology can be generated from knowledge in various languages. Regardless of the language used, users must process the corpus in that language and refine key fields for ontology generation. Some languages, such as Chinese, have no clear delimiter between words, which is more difficult or complicated in terms of language processing compared to English, making keyword extraction more difficult. Therefore, the semantic content of the Chinese text corpus is not easy to understand. Natural language processing (NLP) and latent semantic analysis (LSA) are used in computer science in areas involving interaction between computers and human languages. Combining NLP and LSA can perform lexical, grammatical, syntactic, and semantic analysis on Chinese text corpora. This kind of analysis especially involves word segmentation, part-of-speech tagging, word extraction, statistical analysis, and determination of word relevance. However, due to the complexity of the Chinese language, NLP and LSA may not necessarily effectively and accurately refine the correct keywords or concepts for ontology generation.

In short, there is a need for a more efficient system and method. Ideally, a computer-implemented method and system is needed for the generation of Chinese ontology libraries to meet the rapidly developing data world and meet the needs of data users.

The use of structured network knowledge can automatically construct the Chinese ontology library. Structured network knowledge is a database of structured messages stored on the network. For example, there are many web-based Chinese encyclopedias, such as Baidu Encyclopedia and Chinese Wikipedia, which are popular public knowledge bases consisting of millions of articles. Each article contains a topic, which is usually manually edited by a data user with knowledge of the topic. If you find an error or invalid message, you can Lu's encyclopedia reports to correct those errors or invalid messages. Therefore, each topic can be considered to be edited manually and selected by an expert, and therefore can be considered to be an expert opinion on the topic. Each topic can be further treated as a concept when used to generate an ontology. In addition, data users can display related articles by inserting links into the articles. This kind of link can be considered as the connection point in the concept, and thus represents the semantic relationship between different concepts. Because structured network knowledge is based on the inclusion of a large number of concepts and the relationships between concepts, unlike ANN, which requires pre-training, the generation of ontology using structured network knowledge can be done automatically without a lot of human preparation data. Therefore, the present invention does not require any human intervention and is therefore more efficient in terms of ontology generation.

Since the Chinese language has no clear delimiters between words, the accuracy of generating the extracted knowledge in the Chinese ontology database usually depends on the way of sentence segmentation and which word examples are selected for extraction. To generate Chinese ontology libraries, NLP and LSA are usually used for knowledge extraction. NLP and LSA are computer-executable programs that perform lexical, grammatical, syntactic, and semantic analysis of Chinese text corpora. NLP and LSA can be considered to use computer language to understand human language, and this understanding may not be accurate and effective compared to the understanding of Chinese corpus by native Chinese speakers. With this in mind, the present invention uses hyperlinks in a structured knowledge network to discover related concepts to effectively extract Chinese knowledge. Since these hyperlinks have been reviewed by experts, it can be said that they can more accurately describe the relationship between concepts.

Described below is a method and computer-readable medium for the construction of an automatic Chinese ontology library based on structured network knowledge, and its coding enables the processor to implement The instructions of the method include the following steps: grabbing structured knowledge from a structured knowledge network, where the structured knowledge includes at least one concept for automatic Chinese ontology library generation; filtering irrelevant links; extracting Knowledge related to the aforementioned concept of interest; discovering related concepts of the aforementioned concept of interest; inferring the aforementioned concept of interest and the semantic relevance of its associated concept through a measure of cosine similarity; and storing the inferred aforementioned Semantic relevance data.

Ideally, the steps of structured knowledge captured from a structured knowledge network include the following steps: browsing the aforementioned structured knowledge through a Hypertext Transfer Protocol ("HTTP") protocol; accessing the structured knowledge classification page using a breadth-first search method Link in the Chinese text until all the linked Chinese text corpora are accessed; at least one Chinese text corpus is obtained from the aforementioned structured knowledge network, wherein the subject, abstract, and content of the aforementioned Chinese text corpus consist of the aforementioned Chinese text The HTML header, title, and body tags in the static Hypertext Markup Language ("HTML") page of the corpus are determined; and a link record is generated for each Chinese text corpus obtained.

Further, the step of structured knowledge captured from the structured knowledge network includes the following steps: generating a unique identifier for each Chinese text corpus obtained.

Further, the step of capturing the structured knowledge from the structured knowledge network includes the following steps: storing a URL ("URL"), an identifier, and / or a last modification time for each Chinese text corpus obtained.

Further, the step of capturing the structured knowledge from the structured knowledge network includes the following steps: scanning all the acquired Chinese text corpora at a preset time interval; generating by retrieving whether there are matching records with the same last modification time Or update Chinese text corpus records; and eliminate all duplicate Chinese text corpora.

Further, the steps of eliminating duplicate Chinese text corpora include the following steps: Each Chinese text corpus retains only one identifier; and all other different identifiers of the same Chinese text corpus are converted into redirection identifiers.

Ideally, the steps of filtering irrelevant links include the following steps: irrelevant links to external web pages, irrelevant links in the access menu that do not involve the aforementioned conceptual knowledge of interest, and recurrence in the aforementioned structured knowledge network The links are filtered for noise.

Ideally, the step of extracting knowledge related to the aforementioned concept of interest includes the following steps: extracting relevant noun terms from a Chinese text corpus describing the concept of interest.

Ideally, the step of discovering related concepts of the aforementioned concept of interest includes the following steps: extracting a list of hyperlinks from the Chinese text corpus of the concept of interest, where the Chinese text corpus of each hyperlink represents the same Concept related to the concept of interest.

Ideally, the step of inferring the semantic relevance of the aforementioned concept of interest and its associated concepts through a measure of cosine similarity includes the steps of: calculating the term frequency weight vector V1 of the aforementioned concept of interest; accessing the Chinese of the aforementioned concept of interest Superlinks in the text corpus to locate the associated concepts of the aforementioned concepts of interest; compute the term frequency weight vector for each of the aforementioned associated concepts, where the aforementioned term frequency weight vector for each of the aforementioned associated concepts represents each The unique semantics of the associated concepts; and calculate the cosine similarity between the concept of interest and the term frequency weight vector for each associated concept.

Further, the term frequency weight vector V1 is calculated by the following equation: V1 = (tf (t1, c1), tf (t2, c1), ... tf (tn, c1))

Where tf (t1, c1) is the term frequency of the first related term in the Chinese text corpus of the concept c1 of interest; tf (t2, c1) is the second correlation in the Chinese text corpus of the concept of interest c1 Terminology Frequency; and tf (tn, c1) is the term frequency of the nth related term in the Chinese text corpus of the concept c1 of interest.

Further, the term frequency weight vector for each associated concept is calculated by the following equation: V2 = (tf (t1, c2), tf (t2, c2), ...... tf (tn, c2))

Where V2 is the term frequency weight vector of the related concept c2; tf (t1, c2) is the term frequency of the first related term in the Chinese text corpus of the aforementioned related concept c2; tf (t2, c2) is the aforementioned related The term frequency of the second related term in the Chinese text corpus of the related concept c2; and tf (tn, c2) is the term frequency of the nth related term in the Chinese text corpus of the related concept c2.

In addition, the steps to calculate the cosine similarity between the concept of interest and the term frequency weight vector for each associated concept are:

Where V1 and V2 are the term frequency weight vectors of the concept of interest c1 and the associated concept c2, respectively.

In addition, the step of storing the inferred aforementioned semantic relevance data includes: storing semantic relevance in a web ontology language; and indexing the aforementioned semantic relevance information.

Ideally, the web ontology language used is a resource description framework ("RDF").

Ideally, the step of indexing the aforementioned semantically relevant messages includes establishing Includes concept maps of concepts of interest, related concepts, number of related concepts, and RDF icons.

Ideally, the steps to capture structured knowledge from a structured knowledge network include the following steps: Grab structured knowledge from a web-based Chinese encyclopedia.

Ideally, the step of capturing structured knowledge from a structured knowledge network includes the following steps: Grab structured knowledge from Baidu Encyclopedia or Chinese Wikipedia.

In addition, a system for constructing an automatic Chinese ontology library based on structured network knowledge includes: a web crawling module configured to capture structured knowledge from a structured knowledge network; a noise filtering module configured to filter unrelated links A knowledge extraction module configured to extract knowledge related to the concept of interest in the Chinese text corpus; a database storing Chinese text corpora downloaded from structured network knowledge; and a relationship discovery module configured to extract A related concept of the concept of interest, and a measure of cosine similarity is used to calculate the semantic correlation between the concept of interest and the related concept.

Ideally, the irrelevant link is an irrelevant link connected to an external webpage, an irrelevant link in the access menu that does not involve the aforementioned conceptual knowledge of interest, and a recurring link in the aforementioned structured knowledge network.

In addition, the system includes a visual interface displaying a concept map, wherein the aforementioned concept map includes the concerned concept, related concepts, the number of related concepts, and RDF icons, wherein the number of related concepts is the aforementioned related to the aforementioned concerned concept. For the total number of related concepts, the aforementioned RDF icon allows the user to download the RDF triples of the aforementioned concepts of interest.

Ideally, semantic relevance is encoded by RDF.

The present invention provides a more efficient system and method for automatic Chinese ontology library generation, in order to cope with the rapidly developing digital world and meet the needs of data users.

1‧‧‧ static HTML page

2‧‧‧Automatic Chinese Ontology Library Construction System Based on Structured Network Knowledge

21‧‧‧Web Crawl Module

22‧‧‧Noise Filter Module

23‧‧‧Knowledge Extraction Module

24‧‧‧Database

25‧‧‧ Relationship Discovery Module

26‧‧‧Visualization Module

51‧‧‧Concerned concepts

52‧‧‧related concepts

[Figure 1] A block diagram of a possible implementation of a system for building an automatic Chinese ontology library based on structured network knowledge.

[Figure 2] A flowchart showing the main steps of constructing an automatic Chinese ontology library based on structured network knowledge.

[Fig. 3] A flowchart showing further steps of relationship discovery.

[Figure 4] is a conceptual diagram of the concept "Three Kingdoms".

[Figure 5] is the theme and mutual semantic relevance displayed in RDF format.

The details of the exemplary embodiments are described in detail with reference to the examples shown in the accompanying drawings, wherein like reference numerals refer to like elements throughout.

By way of illustration only, the preferred embodiments are described in the drawings and below. It should be noted that according to the discussion below, alternative embodiments of the structures and methods disclosed herein will undoubtedly be considered as viable alternatives without departing from the principles claimed.

The implementation of the systems, methods, and computer-readable media described herein is based on a structured network knowledge automatic Chinese ontology library construction.

It can be seen from FIG. 1 that the system 2 based on the structured online knowledge automatic Chinese ontology library construction includes a web crawling module 21, a noise filtering module 22, a knowledge extraction module 23, a database 24, Relationship discovery module 25 and visualization module 26. Figure 2 shows a flowchart of the construction of a structured online knowledge-based automatic Chinese ontology library.

In step S21, a static HTML webpage 1 of a structured knowledge network such as a web-based Chinese encyclopedia can be crawled from the web through the web crawling module 21. For example, web-based Chinese encyclopedias can be the well-known Baidu Encyclopedia and Chinese Wikipedia. Each static HTML page 1 describes a specific concept and has links to related web pages. In order to crawl all the static HTML pages 1 (including all linked pages) from the structured knowledge pages, the web crawler module 21 browses the directories in the structured knowledge network through the HTTP protocol, and uses the breadth-first search method to access the directories Hyperlinks in a web page until all linked directories are accessed. The web crawling module 21 then obtains and extracts only the Chinese text corpus from the linked static HTML webpage 1, where the subject, abstract, and content are the HTML tags (such as the header, title, and body) on the acquired static HTML page. Label).

A possible implementation manner of the web crawling module 21 is described below. The web crawling module 21 can use the regular notation "<a (. *?) </ A>" to find all possible links from the structured knowledge network, and establish a link for each acquired Chinese text corpus Make a record, and save the link record and the Chinese text corpus obtained in the database 24. Each Chinese text corpus obtained from the captured static HTML webpage 1 can be identified by the URL of the captured static HTML webpage 1. For easy identification, a unique identifier may be generated for the Chinese text corpus based on a web address ("URL") representing the Chinese text corpus. For example, if a Chinese text corpus A is obtained from the static HTML webpage 1 captured at the URL http://baike.baidu.com/view/2347.htm, then the Chinese text corpus A will have an identifier of 2347. If the static HTML webpage 1 captured from the URL http://baike.baidu.com/view/10088.htm is obtained in Chinese Text corpus B, then the Chinese text corpus B will have an identifier of 10088. The URL, identifier and last modification time of each Chinese text corpus are stored in the database 24.

The web crawler module 21 scans all downloaded Chinese texts at a preset time interval, and establishes or updates by retrieving whether the last modified time of the downloaded Chinese text corpus matches the last modified time in the existing link record. Stored link records. The web crawling module 21 can also scan and find the same Chinese text corpus in two or more captured static HTML webpages 1 with different URLs. For example, the same Chinese text corpus may exist under the browse page and sub-browse page of the crawled static HTML webpage 1 with the following different URLs: (under the browse page) http://baike.baidu.com/view/1005619. htm (under the sub-browsing page) http: //baike.baidu.com/subview/1005619/1005619.htm This copy of Chinese text corpora obtained from different URLs will generate different identifiers and make the identifiers not unique. In order to eliminate the duplicate Chinese text corpus in the database 24, the web crawling module 21 may set the identifier of the Chinese text corpus in the second page as a redirection identifier, and redirect the Chinese text corpus to Browse the identifiers below the page. Therefore, each Chinese text corpus has only one identifier, thereby maintaining the uniqueness of the identifiers in the linked records.

In short, the web crawling module 21 can scan all the link records extracted by the above-mentioned regular representation, extract the identifier from the link by matching the "href" attribute value in the <a> tag, and use the identifier to find The unique identifier stored in the corpus recorded by the database 24 is updated when the link record redirection identifier exists. Next, link records of all downloaded Chinese text corpora are created in the database 24.

In step S22, the noise filtering module 22 filters all the irrelevant links connected to the external webpage, the irrelevant links in the access menu unrelated to the knowledge described in the Chinese text corpus, and Recurring links in a structured network of knowledge.

Each obtained Chinese text corpus can represent a concept, and this concept is often the subject of the Chinese text corpus. Concept is an abstract idea. By examining the detailed information related to the concept, people can understand the concept related to the event, person, object, place, time, characteristics and characteristics and so on. All of the above messages can be considered conceptual knowledge. In step S23, the knowledge extraction module 23 extracts the conceptual knowledge in the Chinese text corpus. There are many ways to extract conceptual knowledge. One method is to extract relevant noun terms from the Chinese text corpus describing the concept. It can be understood that, without departing from the spirit and scope of the present invention, any essentially accurate knowledge extraction measures derived from all known or future-developed means can be taken.

The knowledge extracted from the Chinese text corpus can be used to calculate the term frequency weight vector of the aforementioned Chinese text corpus. Since each Chinese text corpus represents a concept, the term frequency weight vector of the Chinese text corpus can also be a term frequency weight vector of a concept. V1 is the term frequency weight vector for the concept c1 of interest and is calculated as follows: V1 = (tf (t1, c1), tf (t2, c1), ... tf (tn, c1))

Where tf (t1, c1) is the term frequency of the first related term in the Chinese text corpus of the concept c1 of interest; tf (t2, c1) is the second correlation in the Chinese text corpus of the concept of interest c1 The term frequency of terms; and tf (tn, c1) is the term frequency of the nth related term in the Chinese text corpus of the concept c1 of interest.

Chinese text corpora has hyperlinks to other Chinese text corpora. This These hyperlinked Chinese text corpora represent concepts associated with the original concept of interest. In step S24, the relationship discovery module 25 calculates the term frequency weight vectors obtained on the Chinese text corpus (representing the concept of interest) and the hyperlink text corpus (representing the associated concept), and calculates the Chinese text corpus And the cosine similarity of the frequency weight vector of the hypertext Chinese corpus term to discover the connection between the concepts.

As further illustrated in FIG. 3, a possible implementation manner of the relationship discovery module 25 is described as follows. In step S31, a step of extracting a hyperlink list from the captured static HTML webpage 1 of the concept c1 is performed. Each hyperlink in the Chinese text corpus represents an associated concept. In step S32, the related concepts are identified by accessing the hyperlinks found in the Chinese text corpus of the concept of interest. The corresponding term frequency weight vector for associated concepts can also be found. For example, the related concepts c2 and c3 can be found in the Chinese text corpus of the concept c1 of interest, and the term frequency weight vectors of the related concepts c2 and c3 can be calculated as follows: V2 = (tf (t1, c2), tf (t2, c2), ...... tf (tn, c2))

V3 = (tf (t1, c3), tf (t2, c3), ... tf (tn, c3)) where V2 is the term frequency weight vector of the associated concept c2; V3 is the term frequency of the associated concept c3 Weight vector; tf (t1, c2) is the term frequency of the first related term in the Chinese text corpus of the associated concept c2; tf (t2, c2) is the second term in the Chinese text corpus of the associated concept c2 Term frequency of related terms; and tf (tn, c2) is the term frequency of the nth related term in the Chinese text corpus of the related concept c2; tf (t1, c3) is the Chinese text of the related concept c3 Term frequency of the first related term in the data; tf (t2, c3) is the term frequency of the second related term in the Chinese text corpus of the related concept c3; and tf (tn, c3) is the related concept The term frequency of the n-th related term in the Chinese text corpus of c3; in step S33, each associated concept has a term frequency weight vector representing its unique semantics. In step S34, the semantic relevance of the associated concepts is inferred from the cosine similarity measure. The similarity between the two concepts can be inferred from the cosine similarity of a concept and its associated concepts, that is, to measure the cosine angle of the term frequency weight vector of a concept and associated concepts:

Where V1 and V2 are the term frequency weight vectors of the concept of interest c1 and the associated concept c2, respectively.

If the cosine similarity between two concepts is close to 1, then the content between the two concepts is largely similar to each other. In other words, these two concepts may be largely semantically related. If the cosine similarity between two concepts is equal to zero, then the two concepts have completely different content, meaning that they may be completely irrelevant from a semantic point of view. Therefore cosine similarity helps to quantify the similarity of related concepts.

All Chinese text corpus records can be obtained from the database 24, each of which represents a concept, and the term frequency weight vector of each Chinese text corpus is calculated. Derive a record of each Chinese text corpus record and all Chinese text corpus records connected to it through a hyperlink Cosine similarity between. The main body can be coded in a formal language, such as the web ontology language "OWL", the resource description framework ("RDF" or "RDFS"). Other ontology languages can also be used. In this embodiment, as shown in FIG. 5, the Chinese text corpus is converted into RDF triples. All related concepts with the term frequency weighting are also recorded as RDF triples. For example, all related concepts of a Chinese text corpus with semantic relevance are stored in RDF format at step S35, and an index of RDF files with semantic relevance information is established at step S36. The generated RDF triples and stored RDF data can be used for further queries and operations.

In order to facilitate the retrieval of concepts when generating the Chinese ontology library, indexes of titles and abstracts can be established. Conceptual retrieval can be achieved by measuring the relevance of concepts and showing the associated concepts in the concept map.

In one embodiment, in the form of a concept map user interface as shown in FIG. 4, the system 2 includes a visual interface 26 to facilitate the search. The visual interface 26 shows a concept map, in which the concerned concepts 51 (referred to as “three kingdoms” in this embodiment) are displayed in the center of the figure, and all associated concepts 52 are displayed in the periphery. A number under the concept of interest 51 represents the total number of concepts 52 associated with the concept of interest 51. As shown in Figure 4, there are 707 concepts associated with the "Three Kingdoms." The visualization interface 26 can also display RDF icons, allowing users to download RDF triples of the concept 51 of interest. Without departing from the scope of the present disclosure, the concepts and associated concepts, the position and direction of the number of RDF icons may vary.

Specific reference is made herein to the description and examples of exemplary embodiments, but it will be understood that variations and modifications within the spirit and scope of the scope of the patent application are also valid. The above specific embodiments show possible scopes of the description, but are not limited to the scope of the disclosure.

1‧‧‧ static HTML page

2‧‧‧Automatic Chinese Ontology Library Construction System Based on Structured Network Knowledge

21‧‧‧Web Crawl Module

22‧‧‧Noise Filter Module

23‧‧‧Knowledge Extraction Module

24‧‧‧Database

25‧‧‧ Relationship Discovery Module

26‧‧‧Visualization Module

Claims (42)

A method for constructing an automatic Chinese ontology library based on structured network knowledge, which is characterized by the following steps: structured knowledge captured from a structured knowledge network, wherein the structured knowledge includes at least one concept of interest Used for automatic Chinese ontology library construction; filtering irrelevant links; extracting knowledge related to the aforementioned concepts of interest; discovering related concepts of the aforementioned concepts of interest; inferring the aforementioned concepts of interest through measures of cosine similarity and The semantic relevance of its associated concepts; and stores the aforementioned semantic relevance data inferred. The method as described in item 1 of the patent application scope, wherein the step of capturing structured knowledge from the structured knowledge network includes the following steps: using the Hypertext Transfer Protocol ("HTTP") protocol of the aforementioned structured knowledge Browse; use the breadth-first search method to access the category pages of structured knowledge until access to all linked Chinese text corpora; obtain at least one Chinese text corpus from the aforementioned structured knowledge network, where the subject of the aforementioned Chinese text corpus, The abstract and content are determined by the HTML header, title, and body tags displayed on the static Hypertext Markup Language ("HTML") page containing the aforementioned Chinese text corpus; and a link record is generated for each Chinese text corpus obtained . The method described in item 2 of the scope of patent application, further comprising the following steps: generating a unique identifier for each Chinese text corpus obtained. The method described in item 3 of the scope of patent application, further comprising the following steps: storing a Chinese website ("URL"), an identifier, and / or a last modification time for each Chinese text corpus obtained. The method as described in item 4 of the scope of patent application, further comprising the steps of: scanning all acquired Chinese text corpora at a preset time interval; generating or retrieving whether there are matching records with the same last modification time or Update Chinese text corpus records; and eliminate all duplicate Chinese text corpora. The method as described in item 5 of the scope of patent application, wherein the foregoing step of eliminating all duplicate Chinese text corpora includes the following steps: retaining only one identifier for each Chinese text corpus; and allocating the same Chinese text corpus Other different identifiers are converted into redirection identifiers. The method as described in item 1 of the scope of patent application, wherein the step of filtering the irrelevant links includes the following steps: irrelevant links connected to external web pages, and irrelevant links that do not involve the concept knowledge concerned in the access menu , And repetitive links in the aforementioned structured knowledge network for noise filtering. The method described in item 1 of the scope of patent application, wherein the aforementioned step of extracting knowledge related to the aforementioned concept of interest includes the following steps: extracting relevant noun terms from a Chinese text corpus describing the concept of interest. The method as described in item 1 of the scope of patent application, wherein the step of discovering related concepts of the aforementioned concept of interest includes the following steps: extracting from the Chinese text corpus of the concept of interest A list of hyperlinks, where the Chinese text corpus of each hyperlink represents a concept related to the aforementioned concept of interest. The method as described in item 1 of the scope of patent application, wherein the step of inferring the semantic relevance of the aforementioned concept of interest and its related concepts through the measurement of cosine similarity includes the following steps: calculating the terms of the aforementioned concept of interest Frequency weight vector V1; access the hyperlinks in the Chinese text corpus of the aforementioned concept of interest to locate the associated concepts of the aforementioned concept of interest; calculate the term frequency weight vector of each of the aforementioned associated concepts, where each of the aforementioned The aforementioned term frequency weight vector of associated concepts represents the unique semantics of each associated concept; and the cosine similarity between the concept of interest and the term frequency weight vector of each associated concept is calculated. The method described in item 10 of the scope of patent application, wherein the foregoing step of calculating the term frequency weight vector V1 is implemented by the following equation: V1 = (tf (t1, c1), tf (t2, c1), ... tf (tn, c1)) where tf (t1, c1) is the term frequency of the first related term in the Chinese text corpus of the concept c1 of interest; tf (t2, c1) is the Chinese text of the concept c1 of interest The term frequency of the second related term in the data; and tf (tn, c1) is the term frequency of the nth related term in the Chinese text corpus of the concept c1 of interest. The method as described in claim 10, wherein the term of each associated concept is frequently The step of the rate weight vector is achieved by the following equation: V2 = (tf (t1, c2), tf (t2, c2), ... tf (tn, c2)) where V2 is the term frequency weight of the associated concept c2 Vector; tf (t1, c2) is the term frequency of the first related term in the Chinese text corpus of the aforementioned related concept c2; tf (t2, c2) is the first term in the Chinese text corpus of the aforementioned related concept c2 The term frequency of the two related terms; and tf (tn, c2) is the term frequency of the nth related term in the Chinese text corpus of the aforementioned related concept c2. The method as described in item 10 of the scope of patent application, wherein the step of calculating the cosine similarity between the concept of interest and the term frequency weight vector of each associated concept is:

Where V1 and V2 are the term frequency weight vectors of the concept c1 of interest and the related concept c2, respectively. The method described in item 1 of the scope of patent application, wherein the step of storing the inferred aforementioned semantic relevance data includes: storing the semantic relevance in a web ontology language; and indexing the aforementioned semantic relevance information. The method described in item 14 of the scope of patent application, wherein the aforementioned web ontology language is a resource description framework ("RDF"). The method described in item 14 of the scope of patent application, wherein the aforementioned step of indexing the aforementioned semantic relevance information includes: establishing a concept map including the concept of interest, related concepts, the number of related concepts, and the RDF icon. The method described in item 1 of the scope of patent application, wherein the aforementioned step of capturing structured knowledge from the structured knowledge network includes the following steps: structured knowledge captured from a web-based Chinese encyclopedia. The method described in item 1 of the scope of patent application, wherein the aforementioned step of capturing structured knowledge from the structured knowledge network includes the following steps: capturing structured knowledge from Baidu Encyclopedia or Chinese Wikipedia. A system for constructing an automatic Chinese ontology library based on structured network knowledge, which includes: a web crawling module configured to capture structured knowledge from a structured knowledge network; a noise filtering module configured to filter unrelated chains A knowledge extraction module configured to extract knowledge related to the concept of interest in the Chinese text corpus; a database storing Chinese text corpora downloaded from structured network knowledge; and a relationship discovery module configured to The related concepts of the concept of interest are extracted, and the semantic correlation between the concept of interest and the related concept is calculated using the measure of cosine similarity. The system described in claim 19, wherein the irrelevant link is an irrelevant link connected to an external web page, an irrelevant link in the access menu that does not involve the aforementioned conceptual knowledge of interest, and the aforementioned structured knowledge Repeated links on the web. The system described in item 19 of the scope of patent application, further comprising a visual interface displaying a concept map, wherein the aforementioned concept map includes the concept of interest, related concepts, the number of related concepts, and RDF icons. The system according to item 21 of the scope of patent application, wherein the number of the aforementioned related concepts is the total number of the aforementioned related concepts related to the aforementioned concerned concept. The system described in item 21 of the patent application scope, wherein the aforementioned RDF icon allows the user to download an RDF triple of the aforementioned concept of interest. The system described in claim 19, wherein the aforementioned semantic relevance is encoded by RDF. A computer-readable medium whose characteristics are instructions that enable a processor to implement a method when the processor executes, the method includes the following steps: grabbing structured knowledge from a structured knowledge network, where the structured knowledge Including at least one concept of interest for automatic Chinese ontology library construction; filtering irrelevant links; extracting knowledge related to the aforementioned concept of interest; discovering related concepts of the aforementioned concept of interest; inferring through the measurement of cosine similarity The semantic relevance of the aforementioned concerned concept and its associated concepts is derived; and the inferred aforementioned semantic relevance data is stored. The computer-readable medium as described in claim 25, wherein the step of capturing the structured knowledge from the structured knowledge network includes the following steps: the hypertext transfer protocol ("HTTP" ) Protocol to browse; use the breadth-first search method to access the category pages of structured knowledge until access to all linked Chinese text corpora; obtain at least one Chinese text corpus from the aforementioned structured knowledge network, where the aforementioned Chinese text corpus The theme, abstract, and content of the HTML are determined by the HTML header, title, and body tags displayed on a static Hypertext Markup Language ("HTML") page containing the aforementioned Chinese text corpus; and A link record is generated for each Chinese text corpus obtained. The computer-readable medium as described in item 26 of the patent application scope, wherein the method further includes the following steps: generating a unique identifier for each Chinese text corpus obtained. The computer-readable medium according to item 27 of the scope of the patent application, wherein the method further includes the following steps: storing a Chinese website ("URL"), an identifier, and / or a last modification time for each Chinese text corpus obtained. The computer-readable medium described in item 28 of the scope of patent application, wherein the aforementioned method further includes the following steps: scanning all the acquired Chinese text corpora at a preset time interval; and retrieving whether there are any Match records to generate or update Chinese text corpus records; and eliminate all duplicate Chinese text corpora. The computer-readable medium according to item 29 of the scope of patent application, wherein the aforementioned step of converting a copy of the Chinese text corpus includes the following steps: retaining only one identifier for each Chinese text corpus; and the same Chinese text corpus All other different identifiers are converted into redirection identifiers. The computer-readable medium as described in claim 25, wherein the step of filtering the irrelevant links includes the following steps: for irrelevant links connected to external web pages, access to menus that do not involve the aforementioned conceptual knowledge of interest Irrelevant links and recurring links in the aforementioned structured knowledge network are noisy filter. The computer-readable medium as described in claim 25, wherein the aforementioned step of extracting knowledge related to the aforementioned concept of interest includes the following steps: extracting relevant noun terms from a Chinese text corpus describing the concept of interest. The computer-readable medium as described in claim 25, wherein the step of discovering related concepts of the aforementioned concept of interest includes the following steps: extracting a hyperlink list from the Chinese text corpus of the concept of interest, The Chinese text corpus of each of the hyperlinks represents concepts related to the aforementioned concepts of interest. The computer-readable medium according to item 25 of the scope of patent application, wherein the step of inferring the aforementioned concept of interest and the semantic relevance of its associated concept through the measurement of cosine similarity includes the following steps: calculating the aforementioned interest The term frequency weight vector V1 of the concept; access the super link in the Chinese text corpus of the aforementioned concept of interest to locate the related concepts of the aforementioned concept of interest; calculate the term frequency weight vector of each of the aforementioned concepts, where The aforementioned term frequency weight vector of each aforementioned associated concept represents the unique semantics of each associated concept; and the cosine similarity between the concept of interest and the term frequency weight vector of each associated concept is calculated. The computer-readable medium according to item 34 of the scope of patent application, wherein the foregoing step of calculating the term frequency weight vector V1 is implemented by the following equation: V1 = (tf (t1, c1), tf (t2, c1) ,. ... tf (tn, c1)) where tf (t1, c1) is the first related term in the Chinese text corpus of the concept c1 of interest Language frequency; tf (t2, c1) is the term frequency of the second related term in the Chinese text corpus of the concept c1 of interest; and tf (tn, c1) is the first term in the Chinese text corpus of the concept c1 of interest The term frequency of n related terms. The computer-readable medium as described in item 34 of the scope of patent application, wherein the step of the term frequency weight vector for each associated concept is implemented by the following equation: V2 = (tf (t1, c2), tf (t2, c2 ), ...... tf (tn, c2)) where V2 is the term frequency weight vector of the related concept c2; tf (t1, c2) is the first related term in the Chinese text corpus of the aforementioned related concept c2 Term frequency; tf (t2, c2) is the term frequency of the second related term in the Chinese text corpus of the aforementioned related concept c2; and tf (tn, c2) is the Chinese text corpus of the aforementioned related concept c2 The term frequency of the nth related term in. The computer-readable medium as described in claim 34, wherein the steps of calculating the cosine similarity between the concept of interest and the term frequency weight vector of each associated concept are:

Among them, V1 and V2 are the term frequency weight vectors of the concept of interest c1 and the related concept c2, respectively. The computer-readable medium as described in claim 25, wherein the step of storing the inferred semantic relevance data includes: Store the semantic relevance in the web ontology language; and index the aforementioned semantic relevance information. The computer-readable medium as described in item 38 of the scope of patent application, wherein the aforementioned network ontology language is a resource description framework ("RDF"). The computer-readable medium as described in claim 38, wherein the step of indexing the aforementioned semantic relevance information includes: establishing a concept including the concept of interest, related concepts, the number of related concepts, and the RDF icon. Concept map. The computer-readable medium as described in claim 25, wherein the aforementioned step of capturing structured knowledge from the structured knowledge network includes the following steps: Structured knowledge captured from a web-based Chinese encyclopedia. The computer-readable medium according to item 25 of the patent application scope, wherein the aforementioned step of capturing structured knowledge from the structured knowledge network includes the following steps: capturing structured knowledge from Baidu Encyclopedia or Chinese Wikipedia.