CN110888946A - Entity linking method based on knowledge-driven query - Google Patents
Entity linking method based on knowledge-driven query Download PDFInfo
- Publication number
- CN110888946A CN110888946A CN201911236844.5A CN201911236844A CN110888946A CN 110888946 A CN110888946 A CN 110888946A CN 201911236844 A CN201911236844 A CN 201911236844A CN 110888946 A CN110888946 A CN 110888946A
- Authority
- CN
- China
- Prior art keywords
- entity
- item
- knowledge
- linking
- context
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000005065 mining Methods 0.000 claims abstract description 12
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000003058 natural language processing Methods 0.000 claims description 3
- 230000001788 irregular Effects 0.000 abstract description 3
- 238000004364 calculation method Methods 0.000 description 12
- 230000006872 improvement Effects 0.000 description 9
- 238000012545 processing Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000000116 mitigating effect Effects 0.000 description 1
- 229940028444 muse Drugs 0.000 description 1
- GMVPRGQOIOIIMI-DWKJAMRDSA-N prostaglandin E1 Chemical compound CCCCC[C@H](O)\C=C\[C@H]1[C@H](O)CC(=O)[C@@H]1CCCCCCC(O)=O GMVPRGQOIOIIMI-DWKJAMRDSA-N 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/288—Entity relationship models
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention belongs to the technical field of entity linking, and particularly relates to an entity linking method based on knowledge-driven query, which comprises the following steps of firstly, identifying a named entity designation item in a query sentence of a user based on syntactic analysis; secondly, based on incremental evidence mining, information expansion is carried out on the entity nominal items and the local knowledge base through an external knowledge source; and thirdly, linking the entity nominal items by adopting an inference link algorithm. The invention can solve the problems of context lack, irregular description and the like of the query sentences of the user, reduce the dependency on the local knowledge base and accurately realize the generation and the judgment of the candidate entities, thereby improving the performance of entity link.
Description
Technical Field
The invention belongs to the technical field of entity linking, and particularly relates to an entity linking method based on knowledge-driven query.
Background
As dialog interfaces in web applications become more popular, the interaction becomes more similar to natural language dialog, making natural language understanding a key issue. Deep semantic understanding is essential to improve the accuracy, context and personalization of information exchange in pervasive computing devices through natural language. Entity disambiguation research is carried out on user query data, so that the method is beneficial to accurately understanding the real search intention of a user, and lays a foundation for realizing semantic search. In addition, the user query data contains entity nominal items with rich categories, and powerful entity support is provided for the construction and the updating of the knowledge graph.
However, compared with text data such as news and blogs, the constructors of the user query sentence are masses, and the quality is uneven due to problems such as learning and habits. The user can easily have various problems of word order reversal, spelling error, multi-language fusion, name deformation and the like when inquiring sentences and writing randomly. In addition, the word number of the query is limited by the search engine, so that the query statement is concise in description and lacks context information. These characteristics make the entity linking method suitable for long text unable to be directly applied to user query sentence. An indispensable module in the entity link system is used for searching entities in a knowledge base and generating a candidate entity list for entity nominal items, and the adopted method is mainly name matching, but the name similarity between the entity nominal items and the entities in the knowledge base is reduced due to the problem of name non-specification, so that the real target link entities are omitted. In addition, the candidate entity ranking module needs to measure the correlation between the entity nominal item and the candidate entity by using the context information, and the problem of lack of context of the user query can reduce the context similarity between the entity nominal item and the target link entity, so that the accurate judgment of the candidate entity cannot be realized. For example, given a user query sentence "juxtapose video", where the entity in the real world referred to by the entity designation term "juxtapose" is a farmer singer "zhuying", it is obvious that their names are completely different, and no effective identification information is provided in the query sentence, the link between "juxtapose" and "juju zhuyi" cannot be completed only according to the user query sentence.
The inventor finds that the existing entity link method facing the user query has the following defects: the problems of lack of context and non-standard description of the user query statement exist.
Disclosure of Invention
The invention aims to: aiming at the defects of the prior art, the entity linking method based on knowledge-driven query is provided, the problems of context shortage, non-standard description and the like of query sentences of a user can be solved, the dependency on a local knowledge base is reduced, and the generation and the judgment of candidate entities are accurately realized, so that the entity linking performance is improved.
In order to achieve the purpose, the invention adopts the following technical scheme:
an entity linking method based on knowledge-driven query comprises the following steps:
step one, identifying a named entity designation item in a user query statement based on syntactic analysis;
secondly, based on incremental evidence mining, information expansion is carried out on the entity nominal items and the local knowledge base through an external knowledge source;
and thirdly, linking the entity nominal items by adopting an inference link algorithm.
It should be noted that in the entity linking method of the present invention, firstly, a heuristic method is formulated based on syntactic analysis to identify the named entity referent in the user query sentence, so that a small amount of deep and shallow syntactic knowledge is incorporated, the influence caused by word segmentation errors is alleviated, and the accuracy and integrity of the identification of the entity referent are improved; secondly, aiming at the problems of lack of context of user query sentences, non-standard description of entity nominal items, imperfect knowledge of entities in a local knowledge base and the like, information expansion is carried out on the entity nominal items and the local knowledge base by virtue of external knowledge sources such as Baidu search, Baidu encyclopedia and the like on the basis of the idea of incremental evidence mining, so that candidate entities can be generated and judged more accurately; finally, through a reasoning linkage algorithm, the algorithm achieves the aim of improving the entity linkage performance without sacrificing the calculation cost by comprehensively considering and gradually utilizing the knowledge of the entity in various aspects.
As an improvement of the entity linking method based on knowledge-driven query according to the present invention, in step three, the inference linking algorithm includes: generating a candidate entity; ranking the candidate entities; null link designations are predicted.
As an improvement of the entity linking method based on knowledge-driven query according to the present invention, in step three, the inference linking algorithm includes:
and determining the target link entity of the entity named item through the name similarity, the category consistency, the context similarity and the semantic correlation between the entities.
As an improvement of the entity linking method based on knowledge-driven query according to the present invention, in step three, the inference linking algorithm includes:
acquiring the entity designated item from the context information of the entity designated item;
acquiring a target link entity of the entity named item, and then acquiring an associated entity of the target link entity according to an associated structure between entities in the local knowledge base;
and measuring the correlation between the entity designated item and the associated entity by comprehensively utilizing the name, the context similarity and the category consistency.
As an improvement of the entity linking method based on knowledge-driven query, the method for measuring the correlation between the entity nominal item and the associated entity comprises the following steps:
and if the overall similarity score exceeds a threshold value, the associated entity is the target link entity, otherwise, the entity index item is a null link entity index item.
As an improvement of the entity linking method based on knowledge-driven query according to the present invention, in the first step, identifying the entity designation includes:
using a HanLP natural language processing tool to perform word segmentation, part-of-speech tagging and dependency analysis on the input text d to obtain a word set with part-of-speech tags and dependency tags
Wherein, wiRepresents said word, piRepresents said part of speech tag, diRepresenting the dependency label.
As an improvement of the entity linking method based on knowledge-driven query according to the present invention, in the second step, the incremental evidence mining includes:
and associating the entity designation item and the entity in the local knowledge base with the external knowledge source vocabulary by using the entity name, the context and the popularity.
As an improvement of the entity linking method based on knowledge-driven query according to the present invention, in the second step, the method further includes:
and if the entity nominal item is not associated with an external knowledge source entry, putting the user query statement into a search engine, and then acquiring nominal words from the title and the abstract of a search result as the context of the entity nominal item.
As an improvement of the entity linking method based on knowledge-driven query according to the present invention, in the second step, information expansion is performed on the entity named item and the local knowledge base, including
Expanding the alias, the category and the context of the entity named item through an external knowledge source;
and acquiring entity aliases, abundant categories and extended description information, and optimizing the local knowledge base.
The method has the advantages that the method comprises the following steps of firstly, identifying the named entity designation item in the query sentence of the user based on syntactic analysis; secondly, based on incremental evidence mining, information expansion is carried out on the entity nominal items and the local knowledge base through an external knowledge source; and thirdly, linking the entity nominal items by adopting an inference link algorithm. In the entity linking method, firstly, a heuristic method is formulated based on syntactic analysis to identify the named entity nomination items in the user query sentence, so that a small amount of deep and shallow syntactic knowledge is integrated, the influence caused by word segmentation errors is relieved, and the accuracy and the integrity of the entity nomination item identification are improved; secondly, aiming at the problems of lack of context of user query sentences, non-standard description of entity nominal items, imperfect knowledge of entities in a local knowledge base and the like, information expansion is carried out on the entity nominal items and the local knowledge base by virtue of external knowledge sources such as Baidu search, Baidu encyclopedia and the like on the basis of the idea of incremental evidence mining, so that candidate entities can be generated and judged more accurately; finally, through a reasoning linkage algorithm, the algorithm achieves the aim of improving the entity linkage performance without sacrificing the calculation cost by comprehensively considering and gradually utilizing the knowledge of the entity in various aspects. The invention can solve the problems of context lack, irregular description and the like of the query sentences of the user, reduce the dependency on the local knowledge base and accurately realize the generation and the judgment of the candidate entities, thereby improving the performance of entity link.
Drawings
FIG. 1 is a schematic diagram of a user query entity link according to the present invention.
FIG. 2 is a diagram illustrating dependency resolution of a user query statement according to the present invention.
Detailed Description
As used in the specification and in the claims, certain terms are used to refer to particular components. As one skilled in the art will appreciate, manufacturers may refer to a component by different names. This specification and claims do not intend to distinguish between components that differ in name but not function. In the following description and in the claims, the terms "include" and "comprise" are used in an open-ended fashion, and thus should be interpreted to mean "include, but not limited to. "substantially" means within an acceptable error range, and a person skilled in the art can solve the technical problem within a certain error range to substantially achieve the technical effect.
In the description of the present invention, it is to be understood that the terms "upper", "lower", "front", "rear", "left", "right", horizontal ", and the like indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience in describing the present invention and simplifying the description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention.
In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
The present invention will be described in further detail with reference to fig. 1 to 2, but the present invention is not limited thereto.
An entity linking method based on knowledge-driven query comprises the following steps:
step one, identifying a named entity designation item in a user query statement based on syntactic analysis;
step two, based on incremental evidence mining, performing information expansion on the entity nominal items and the local knowledge base through an external knowledge source;
and step three, adopting a reasoning and linking algorithm to link the entity nominal items.
It should be noted that in the entity linking method of the present invention, firstly, a heuristic method is formulated based on syntactic analysis to identify the named entity referent in the user query sentence, so that a small amount of deep and shallow syntactic knowledge is incorporated, the influence caused by word segmentation errors is alleviated, and the accuracy and integrity of the identification of the entity referent are improved; secondly, aiming at the problems of lack of context of user query sentences, non-standard description of entity nominal items, imperfect knowledge of entities in a local knowledge base and the like, information expansion is carried out on the entity nominal items and the local knowledge base by virtue of external knowledge sources such as Baidu search, Baidu encyclopedia and the like on the basis of the idea of incremental evidence mining, so that candidate entities can be generated and judged more accurately; finally, through a reasoning linkage algorithm, the algorithm achieves the aim of improving the entity linkage performance without sacrificing the calculation cost by comprehensively considering and gradually utilizing the knowledge of the entity in various aspects.
Preferably, in step three, the inference linking algorithm includes: generating a candidate entity; sorting the candidate entities; null link designations are predicted.
Preferably, in step three, the inference linking algorithm includes: generating a candidate entity; sorting the candidate entities; null link designations are predicted. And the reasoning link algorithm comprehensively considers and gradually utilizes the name similarity, the category consistency, the context similarity and the semantic correlation among the entities to define a target link entity of the entity index, and is divided into three parts of generating a candidate entity, sequencing the candidate entity and predicting the empty link index according to the processing flow of entity link.
Preferably, in step three, the inference linking algorithm includes:
and determining the target link entity of the entity named item through the name similarity, the category consistency, the context similarity and the semantic correlation between the entities.
It should be noted that: given entity term miE.g. M, and entity e in local knowledge basejE, by metric miAnd ejTo generate miCandidate entity list Ci. In order to improve the probability of the target link entity contained in the candidate entity list and control the scale of the list, name fuzzy matching and category consistency approximation are adoptedBundle, similarity of name over threshold, and category with miUniform ejAs candidate entities. Wherein, the name similarity adopts a formula
And the calculation is helpful for processing the deformed names such as short names, abbreviations and word order reversal, and the probability of successful matching with the target link entity is improved. The category consistency constraint is used for removing candidate entities with different categories to reduce the scale of the candidate list, reduce the noise of subsequent calculation and improve the processing efficiency. Through the above calculation, if the candidate entity list only contains one candidate entity, it is considered as miIf the candidate entity list is empty, then m is considered to beiAn entry is referred to for a null link.
Preferably, in step three, the inference linking algorithm includes:
acquiring entity designated items from the context information of the entity designated items;
acquiring a target link entity of the entity named item, and then acquiring an associated entity of the target link entity according to an associated structure between entities in a local knowledge base;
and measuring the correlation between the entity nominal items and the associated entities by comprehensively utilizing the names, the context similarity and the category consistency.
Preferably, measuring the correlation between the entity named item and the associated entity comprises:
and if the overall similarity score exceeds the threshold value, the associated entity is the target link entity, otherwise, the entity index item is the null link entity index item.
It should be noted that: candidate entity list C of size greater than 1, by measuring entity designation miAnd its candidate entity eijE with the highest score is selectedijAs the link target entity. In order to reduce the calculation cost, a measurement strategy for gradually using name similarity and context similarity is adopted, wherein the name similarity adopts a formula
If and eijIs the maximum and exceeds the threshold value omega, then e is considered to beijFor the target link entity, if the above conditions are not satisfied, adopting a formula
Calculate miAnd eijE if the similarity score exceeds a threshold valueijFor target link entities, if m existsiIf the above two conditions are not met, it is considered to be a null link designation.
However, m is determined based only on name and context similarity, and category consistencyiFor the null link named item, it is too comprehensive, because of the problems of entity name diversity, dissimilar class description text, and lack of context, etc., the above strategies cannot process the named item whose name and context are dissimilar but point to the same entity, in order to improve the accuracy and recall rate of entity link, based on the idea of "having an association relation with the real entity", the semantic correlation between the entities is used to further link the output null link entity named item set F, and the processing flow is as follows: first, from miObtaining the entity reference term t from the context informationijSecond, get tijTarget linking entity e ofijThen, according to the association structure between the entities in the local knowledge base, e is obtainedijAnd finally, comprehensively utilizing the similarity of the name and the context, the category consistency and the metric miAnd ek, if the overall similarity score exceeds a threshold value, considering the ek as a target link entity, otherwise, miDesignating an entry for a null link entity, where c represents miThe category similarity with ek is a constant, if the category descriptions of the two are consistent, c is 0.1, otherwise c is 0, and the value of c is set without affecting the name and context metrics.
Preferably, in the first step, identifying the entity reference item includes:
using HanLP natural language processing tool to perform word segmentation, part-of-speech tagging and dependency relationship on input text dAnalyzing the system to obtain a word set with part-of-speech labels and dependency relationship labels
Wherein, wiRepresents a word, piRepresenting part-of-speech tags, diRepresenting dependency tags.
It should be noted that: then, based on the assumption that the entity term is a noun word, we use information such as dependency relationship and part of speech among sentence components to formulate the following recognition method.
(1) The entity referents must be noun words and must contain one or more proper noun words, i.e., piN. Where n is a general term of part of speech of a proper noun, it means that the part of speech tag starts with n and does not end with n, such as ns.
(2) If the entity reference includes a common noun word, i.e. piN, there must be a dependency relationship of "fixed relation" with the proper noun word, and there must not be an auxiliary word such as "fixed relation" between the two.
By limiting the recognition range of the entity term, the probability that the noun word is the entity term is improved, the influence of segmentation error on one entity term into a plurality of parts is relieved, and the probability that a plurality of entity terms are recognized as one entity term is reduced, given the input sentence "champion of NBA total resolution in 2014? ", the part of speech tagging and dependency resolution results are shown in FIG. 2. According to the identification method established by the user, the obtained entity named items are ' NBA general finals ' in 2014 ', ' NBA ' and ' champions '.
Preferably, in step two, the incremental evidence mining comprises:
and associating the entity designation item and the entity in the local knowledge base with an external knowledge source vocabulary by using the entity name, the context and the popularity.
Preferably, step two further comprises:
if the entity designation item is not associated with an external knowledge source entry, putting a user query sentence into a search engine, and then acquiring nominal words from the title and the abstract of a search result as the context of the entity designation item.
Preferably, in the second step, the information extension is carried out on the entity nominal item and the local knowledge base, and the method comprises the following steps of
Expanding the alias, the category and the context of the entity named item through an external knowledge source;
and acquiring entity aliases, enriching categories and expanding description information, and optimizing the local knowledge base.
It should be noted that: in order to solve the problems of lack of context, non-standard entity name and the like in a user query statement, the invention expands the information of alias names, categories, contexts and the like of entity nominal items by means of an encyclopedia search engine, an encyclopedia and other external knowledge sources based on the idea of incremental evidence mining. In addition, the problems that the local knowledge base has much noise, such as inaccurate entity names and lack of partial entity description information are considered. In order to improve the accuracy of entity linking, the method optimizes the local knowledge base by means of Baidu encyclopedia, and comprises the steps of obtaining entity aliases, enriching categories, expanding description information and the like. In order to correctly associate an entity name item m and an entity e in a local knowledge base with an encyclopedia entry t, the knowledge of entity names, context, popularity and the like is comprehensively utilized, and a calculation method is used, such as a formula P (a, b) ═ Pn(a，b)+α(pc(a，b)+pp(b) Shown in (c). Wherein, the symbol a represents an entity named item m and an entity e in a local knowledge base, and the symbol b represents an encyclopedia entry t. In order to deal with the problems that an entity nominal item with an irregular name comprises the forms of abbreviation, word order reversal and the like, and the identification method of the entity nominal item is imperfect to cause the problem that the name is incomplete or other words are merged, the similarity p of the name is usedn(a, b) mitigating the effects of the above problem by calculating the probability of the maximum number of common characters between two name strings, such as the formula
Shown, where the MCC function is used to calculate the maximum number of common characters between two strings, the min function represents taking the minimum value, and the Len function represents calculating the length of the string. If p isnAnd (a, b) is greater than the similarity threshold omega, then m is considered to be associated with t. Add context similarity pc(a, b) and entity popularity pp(b) For example, the name 'muse' corresponds to three objects in the real world, namely an actor in China, a poem of China, a duhui-fole, α is a discrimination factor, if the thesaurus has synonyms, α is 1, otherwise α is 0, and in order to reduce the calculation cost, the context similarity p is used for reducing the calculation costc(a, B) by measuring the number of words that are similar between the two sets of contextual words | A | and | B |, as in a formula
Shown, entity popularity pp(b) And measuring, namely adopting the idea that the more frequent the access is, the higher the popularity of the entry is, and adopting the access times v of one entry t in the encyclopedia. Since v is a large positive integer and pc(a, b) is a decimal fraction less than 1, which we convert to so as not to affect the measure of context similarity, e.g. formula
As shown. Where | v | represents the number of bits of the access number. For example, given the number of accesses v 54896 for an entry, v 5, the popularity p of the entry tp(b)＝0.554896。
After acquiring encyclopedia entries by adopting the method, names, aliases and category labels are acquired from the description pages of the entries, and noun words are acquired from texts to serve as context information. Because the noun words are more descriptive and descriptive than words of other parts of speech. On one hand, the calculation cost in subsequent processing is reduced, and on the other hand, the noise information in the context similarity calculation process is removed.
Variations and modifications to the above-described embodiments may also occur to those skilled in the art, which fall within the scope of the invention as disclosed and taught herein. Therefore, the present invention is not limited to the above-mentioned embodiments, and any obvious improvement, replacement or modification made by those skilled in the art based on the present invention is within the protection scope of the present invention. Furthermore, although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.
Claims (9)
1. An entity linking method based on knowledge-driven query is characterized by comprising the following steps:
step one, identifying a named entity designation item in a user query statement based on syntactic analysis;
secondly, based on incremental evidence mining, information expansion is carried out on the entity nominal items and the local knowledge base through an external knowledge source;
and thirdly, linking the entity nominal items by adopting an inference link algorithm.
2. The entity linking method based on knowledge-driven query as claimed in claim 1, wherein in step three, the inference linking algorithm comprises: generating a candidate entity; ranking the candidate entities; null link designations are predicted.
3. The entity linking method based on knowledge-driven query as claimed in claim 1, wherein in step three, the inference linking algorithm comprises:
and determining the target link entity of the entity named item through the name similarity, the category consistency, the context similarity and the semantic correlation between the entities.
4. The entity linking method based on knowledge-driven query as claimed in claim 3, wherein in step three, the inference linking algorithm comprises:
acquiring the entity designated item from the context information of the entity designated item;
acquiring a target link entity of the entity named item, and then acquiring an associated entity of the target link entity according to an associated structure between entities in the local knowledge base;
and measuring the correlation between the entity designated item and the associated entity by comprehensively utilizing the name, the context similarity and the category consistency.
5. The method of claim 4, wherein measuring the correlation between the entity referenceand the associated entity comprises:
and if the overall similarity score exceeds a threshold value, the associated entity is the target link entity, otherwise, the entity index item is a null link entity index item.
6. The entity linking method based on knowledge-driven query as claimed in claim 1, wherein in the first step, identifying the entity reference item comprises:
using a HanLP natural language processing tool to perform word segmentation, part-of-speech tagging and dependency analysis on the input text d to obtain a word set with part-of-speech tags and dependency tags
Wherein, wiRepresents said word, piRepresents said part of speech tag, diRepresenting the dependency label.
7. The entity linking method based on knowledge-driven query as claimed in claim 1, wherein in the second step, the incremental evidence mining comprises:
and associating the entity designation item and the entity in the local knowledge base with the external knowledge source vocabulary by using the entity name, the context and the popularity.
8. The entity linking method based on knowledge-driven query as claimed in claim 7, wherein in said second step, further comprising:
and if the entity nominal item is not associated with an external knowledge source entry, putting the user query statement into a search engine, and then acquiring nominal words from the title and the abstract of a search result as the context of the entity nominal item.
9. The method as claimed in claim 1, wherein the second step of performing information expansion on the entity reference item and the local knowledge base comprises
Expanding the alias, the category and the context of the entity named item through an external knowledge source;
and acquiring entity aliases, abundant categories and extended description information, and optimizing the local knowledge base.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911236844.5A CN110888946A (en) | 2019-12-05 | 2019-12-05 | Entity linking method based on knowledge-driven query |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911236844.5A CN110888946A (en) | 2019-12-05 | 2019-12-05 | Entity linking method based on knowledge-driven query |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110888946A true CN110888946A (en) | 2020-03-17 |
Family
ID=69750650
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911236844.5A Pending CN110888946A (en) | 2019-12-05 | 2019-12-05 | Entity linking method based on knowledge-driven query |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110888946A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111563149A (en) * | 2020-04-24 | 2020-08-21 | 西北工业大学 | Entity linking method for Chinese knowledge map question-answering system |
| CN111782769A (en) * | 2020-07-01 | 2020-10-16 | 重庆邮电大学 | Intelligent knowledge graph question-answering method based on relation prediction |
| CN112395429A (en) * | 2020-12-02 | 2021-02-23 | 上海三稻智能科技有限公司 | Method, system and storage medium for determining, pushing and applying HS (high speed coding) codes based on graph neural network |
| CN113569060A (en) * | 2021-09-24 | 2021-10-29 | 中国电子技术标准化研究院 | Standard text based knowledge graph disambiguation method, system, device and medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104933039A (en) * | 2015-06-04 | 2015-09-23 | 中国科学院新疆理化技术研究所 | Entity link system for language lacking resources |
| CN105183770A (en) * | 2015-08-06 | 2015-12-23 | 电子科技大学 | Chinese integrated entity linking method based on graph model |
| CN106960021A (en) * | 2017-03-10 | 2017-07-18 | 北京工业大学 | A kind of enquiry expanding method based on the positive and negative external feedback information of multi-source |
| CN107316062A (en) * | 2017-06-26 | 2017-11-03 | 中国人民解放军国防科学技术大学 | A kind of name entity disambiguation method of improved domain-oriented |
| CN108959258A (en) * | 2018-07-02 | 2018-12-07 | 昆明理工大学 | It is a kind of that entity link method is integrated based on the specific area for indicating to learn |
-
2019
- 2019-12-05 CN CN201911236844.5A patent/CN110888946A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104933039A (en) * | 2015-06-04 | 2015-09-23 | 中国科学院新疆理化技术研究所 | Entity link system for language lacking resources |
| CN105183770A (en) * | 2015-08-06 | 2015-12-23 | 电子科技大学 | Chinese integrated entity linking method based on graph model |
| CN106960021A (en) * | 2017-03-10 | 2017-07-18 | 北京工业大学 | A kind of enquiry expanding method based on the positive and negative external feedback information of multi-source |
| CN107316062A (en) * | 2017-06-26 | 2017-11-03 | 中国人民解放军国防科学技术大学 | A kind of name entity disambiguation method of improved domain-oriented |
| CN108959258A (en) * | 2018-07-02 | 2018-12-07 | 昆明理工大学 | It is a kind of that entity link method is integrated based on the specific area for indicating to learn |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111563149A (en) * | 2020-04-24 | 2020-08-21 | 西北工业大学 | Entity linking method for Chinese knowledge map question-answering system |
| CN111563149B (en) * | 2020-04-24 | 2023-01-31 | 西北工业大学 | Entity linking method for Chinese knowledge map question-answering system |
| CN111782769A (en) * | 2020-07-01 | 2020-10-16 | 重庆邮电大学 | Intelligent knowledge graph question-answering method based on relation prediction |
| CN111782769B (en) * | 2020-07-01 | 2022-07-08 | 重庆邮电大学 | Intelligent knowledge graph question-answering method based on relation prediction |
| CN112395429A (en) * | 2020-12-02 | 2021-02-23 | 上海三稻智能科技有限公司 | Method, system and storage medium for determining, pushing and applying HS (high speed coding) codes based on graph neural network |
| CN113569060A (en) * | 2021-09-24 | 2021-10-29 | 中国电子技术标准化研究院 | Standard text based knowledge graph disambiguation method, system, device and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8073877B2 (en) | Scalable semi-structured named entity detection | |
| RU2375747C2 (en) | Checking relevance between key words and website content | |
| US6687697B2 (en) | System and method for improved string matching under noisy channel conditions | |
| Cohen et al. | Exploiting dictionaries in named entity extraction: combining semi-markov extraction processes and data integration methods | |
| CN110888946A (en) | Entity linking method based on knowledge-driven query | |
| US8606559B2 (en) | Method and apparatus for detecting errors in machine translation using parallel corpus | |
| US7254774B2 (en) | Systems and methods for improved spell checking | |
| KR100630886B1 (en) | Character string identification | |
| US7493251B2 (en) | Using source-channel models for word segmentation | |
| US20080059146A1 (en) | Translation apparatus, translation method and translation program | |
| JP2002510076A (en) | Information retrieval and speech recognition based on language model | |
| JP2002215619A (en) | Translation sentence extracting method from translated document | |
| CN111522910A (en) | Intelligent semantic retrieval method based on cultural relic knowledge graph | |
| US7475005B2 (en) | Translation system, dictionary updating server, translation method, and program and recording medium for use therein | |
| JP5057474B2 (en) | Method and system for calculating competition index between objects | |
| US20090192991A1 (en) | Network information searching method by speech recognition and system for the same | |
| WO2020037794A1 (en) | Index building method for english geographical name, and query method and apparatus therefor | |
| CN115983233A (en) | Electronic medical record duplication rate estimation method based on data stream matching | |
| CN112613293B (en) | Digest generation method, digest generation device, electronic equipment and storage medium | |
| Huang et al. | Keyword spotting in unconstrained handwritten Chinese documents using contextual word model | |
| JP4900947B2 (en) | Abbreviation extraction method, abbreviation extraction apparatus, and program | |
| JP4102153B2 (en) | Post-processing device for character recognition using the Internet | |
| CN111767733A (en) | Document security classification discrimination method based on statistical word segmentation | |
| JP2000259645A (en) | Speech processor and speech data retrieval device | |
| WO2022264435A1 (en) | Device for automatically generating entity, intent, and corpus, and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200317 |
|
| RJ01 | Rejection of invention patent application after publication |