CN112732876B - Universal semantic matching implementation method and system - Google Patents
Universal semantic matching implementation method and system Download PDFInfo
- Publication number
- CN112732876B CN112732876B CN202110179941.6A CN202110179941A CN112732876B CN 112732876 B CN112732876 B CN 112732876B CN 202110179941 A CN202110179941 A CN 202110179941A CN 112732876 B CN112732876 B CN 112732876B
- Authority
- CN
- China
- Prior art keywords
- matching
- semantic
- sub
- rule
- matched
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 21
- 230000014509 gene expression Effects 0.000 claims abstract description 70
- 230000006870 function Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims 1
- 238000010801 machine learning Methods 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 101100489887 Arabidopsis thaliana ABA2 gene Proteins 0.000 description 1
- 101100478320 Caenorhabditis elegans sre-2 gene Proteins 0.000 description 1
- 101000629937 Homo sapiens Translocon-associated protein subunit alpha Proteins 0.000 description 1
- 101000629913 Homo sapiens Translocon-associated protein subunit beta Proteins 0.000 description 1
- 101000697347 Homo sapiens Translocon-associated protein subunit gamma Proteins 0.000 description 1
- 101100501251 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) ELO3 gene Proteins 0.000 description 1
- 102100026231 Translocon-associated protein subunit alpha Human genes 0.000 description 1
- 102100026229 Translocon-associated protein subunit beta Human genes 0.000 description 1
- 102100028160 Translocon-associated protein subunit gamma Human genes 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 101150036683 sre1 gene Proteins 0.000 description 1
- 238000006467 substitution reaction 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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/90335—Query processing
- G06F16/90344—Query processing by using string matching techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/194—Calculation of difference between files
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
Abstract
The invention discloses a universal semantic matching realization method and a system, wherein the method comprises the following steps: s1, acquiring the natural language input by the user and the semantic to be matched; s2, semantic matching, comprising the following steps: s21, searching a semantic rule expression array corresponding to the semantics; s22, traversing each semantic rule expression in the semantic rule expression array; s23, judging whether the natural language input by the user is matched with the semantic rule expression, and judging whether the natural language input by the user is successfully matched with each semantic sub-rule by judging whether the natural language input by the user is successfully matched with each semantic sub-rule and calculating the matching results of a plurality of semantic sub-rules by using a logic calculator, so as to judge whether the whole semantic rule expression is successfully matched, if only one is successfully matched, the natural language input by the user is successfully matched with the semantic, otherwise, the matching is failed; the system comprises: the semantic rule expression array is composed of a plurality of semantic rule expressions, and each semantic rule expression comprises a plurality of semantic sub-rules connected through logical relations.
Description
Technical Field
The invention relates to the technical field of natural language processing, in particular to a universal semantic matching implementation method and a universal semantic matching implementation system.
Background
In the process of natural language processing, semantic matching is an important key link. Semantic matching may generally be understood as the process of determining whether there is a match between the natural language and semantics entered by the user.
Semantic matching is mainly used for solving the diversity, ambiguity and robustness of languages. At present, the commonly used strategy is regular matching and machine learning, but in practical product application, the uncontrollable matching result generated by the machine learning is high, the requirements of users cannot be met in some formal commercial products, and the method is not suitable for serving as the main semantic matching mode of the commercial products. The following are some of the problems faced by current semantic rule matching:
1. multiple expression modes exist in one semantic;
2. one expression may be represented by a different natural language structure;
3. the diversity, ambiguity and robustness of the semantics do not have a universal expression mode to solve the problems, so that the problems of disordered compiling logic, mutual influence between semantics, wrong matching and the like are frequently found by a writer of the semantic rules when the semantic rules are compiled.
Disclosure of Invention
In order to solve the defects of the prior art and realize the purpose of conveniently and flexibly, quickly and accurately defining the matching rules of various semantics by semantic rule writers under the scene of rule matching, the invention adopts the following technical scheme:
a general semantic matching implementation method comprises the following steps:
s1, acquiring the natural language input by the user and the semantic to be matched;
s2, semantic matching, comprising the following steps:
s21, searching a semantic rule expression array SREA corresponding to the semantics;
s22, traversing each semantic regular expression SRE in the semantic regular expression array SREA;
s23, judging whether the natural language input by the user is matched with the semantic rule expression SRE, judging whether the natural language input by the user is successfully matched with each semantic sub-rule SSR, and calculating the matching results of a plurality of semantic sub-rule SSRs by using a logic calculator, thereby judging whether the whole semantic rule expression SRE is successfully matched, if only one SRE is successfully matched, the natural language input by the user is successfully matched with the semantic, otherwise, the matching is failed.
Further, the matching of the semantic sub-rule SSR with the natural language input by the user in the step S23 is represented by a function f (MR, MWA, MA), which includes the following steps:
s231, for all matched sub-elements MSE in the matched phrase MWA, a matching algorithm MA is used for judging whether the natural language input by the user is matched with the matched sub-elements MSE:
when the matching rule MR is all, the matching rule MR calculates that the matching is successful only if all the matching sub-elements MSE are successfully matched;
when the matching rule MR is anyone, as long as any matching sub-element MSE is successfully matched, the matching rule MR is even successfully matched;
when the matching rule MR is none, all the matching subelements MSE can not be successfully matched, and the matching rule MR calculates that the matching is successful;
when the matching rule MR is notall, as long as all the matching sub-elements MSE are not successfully matched, the matching rule MR is even if the matching is successful;
s232, matching algorithm MA:
when the matching algorithm MA is contained, as long as the natural language input by the user contains the matching sub-element MSE, the matching algorithm MA is successful even if the matching is successful;
when the matching algorithm MA is similarity, calculating the similarity between the natural language input by the user and the matching sub-element MSE, and if the similarity exceeds a specified threshold, successfully matching the matching algorithm MA;
when the matching algorithm MA is a regular expression, the regular expression is used for matching the natural language input by the user and the matched sub-element MSE, and the matching algorithm MA is even matched successfully as long as the natural language input by the user and the matched sub-element MSE are matched successfully;
when the matching algorithm is other self-defined matching algorithms, other matching algorithm functions or self-realized matching algorithms can be called to judge whether the matching algorithm is successfully matched.
Further, in step S232, when the matching algorithm MA is similarity, similarity between the natural language input by the user and the sentence as the matching sub-element MSE is calculated by using a text similarity algorithm.
Further, when the matching algorithm MA is a regular expression in step S232, the regular expression is used to match the natural language input by the user and the matching pattern as the matching sub-element MSE.
A general semantic matching implementation system comprises a semantic rule expression array SREA, wherein the semantic rule expression array SREA comprises a plurality of semantic rule expressions SRE, the semantic rule expressions SRE comprise a plurality of semantic sub-rule SSRs, the semantic sub-rule SSRs are connected through a logical relationship, one semantic has a plurality of expression modes, whether a natural language input by a user is matched with the semantic rule expressions SRE or not is judged by traversing each semantic rule expression SRE in the semantic rule expression array SREA, and the natural language input by the user is successfully matched with the semantic rule expression SRE as long as one SRE is successfully matched, otherwise, the matching fails.
Further, the semantic sub-rule SSR comprises a matching rule MR, a matching phrase MWA and a matching algorithm MA;
the matching rule MR is used for matching the rule of the phrase MWA;
the matching phrase MWA comprises matching sub-elements MSE, and has different matching sub-elements MSE under different configuration algorithms MA;
the matching algorithm MA is used for matching the natural language input by the user with the matching sub-element MSE in the matching phrase MWA.
Further, the matching rule MR includes:
all: representing that all matched subelements MSE in the matched phrase MWA must be matched;
and (3) any one of: any one of the matching sub-elements MSE in the matching phrase MWA is represented;
none: indicating that it cannot match any of the matching sub-elements MSE in the matching phrase MWA;
nodall: meaning that it does not match all of the matching sub-elements MSE in the matching phrase MWA.
Further, the matching sub-element MSE is a word, a sentence or a regular matching pattern.
Further, the matching algorithm MA comprises:
comprises the following steps: the semantics of the natural language comprise that a matching sub-element MSE is appointed;
similarity: semantic meaning of natural language and specified matching sub-element MSE, calculating similarity by using a similarity algorithm, and reaching a specified threshold;
the regular expression is as follows: matching the semantics of the natural language with the specified matching sub-element MSE;
self-defining: self-expanding and adding according to the needs.
Further, the logical relationship includes and, or, not, parenthesis.
The invention has the advantages and beneficial effects that:
the invention defines a general grammar written by semantic rules and realizes a corresponding analyzer by a general semantic matching realization method. The method is convenient for semantic rule writers to flexibly, quickly and accurately define the matching rules of various semantics;
2, the method solves the problems of diversity, ambiguity and robustness of semantics, so that the generated matching result has higher controllability and is suitable for commercial products;
3, except for defaulting common matching rules and matching algorithms, the system also reserves an interface which can be added by a user in a self-defined way, so that the system is convenient to expand.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are given by way of illustration and explanation only, not limitation.
A general semantic matching implementation method and system, as shown in fig. 1.
First, noun interpretation:
1. natural Language (Natural Language) input by the user, abbreviation NL;
2. semantics (Semantic), abbreviation S, user expressed intentions, are generally set in advance by data writers and are limited in number. The semantic matching process is a process of judging whether the natural language input by the user is matched with certain semantic;
3. semantic Rule Expression Array (Semantic Rule Expression Array), abbreviated SREA, where a Semantic may have multiple expressions, a Semantic Rule Expression may define an Expression, and multiple expressions may be expressed in the form of an Array, for example: SREA = [ SRE1, SRE 2.., SREn ];
4. semantic Rule Expression (Semantic Rule Expression), SRE, logical Expression composed of Semantic sub-rules, and, or, not, brackets, for example: SRE = SSR1& (SSR2| SSR3) & -SSR 4;
5. semantic Sub-Rule (SSR), abbreviation, contains a matching Rule, a phrase and a matching algorithm, for example: SSR = f (MR, MWA, MA);
(1) matching rules (Match Rule), abbreviated MR, rules for matching phrases, specifically referring to the semantic Rule interpretation part;
(2) matching phrases (Match Word Array), abbreviated MWA, phrases for matching, Sub-elements (Match Sub-elements) thereof, abbreviated MSE, which may be words, sentences or regular matching patterns using different matching algorithm scenarios, etc.;
(3) matching algorithm (Match arithmetric), abbreviated as MA, for matching algorithm of elements in natural language and phrase input by user, specifically referring to semantic rule interpretation part;
6. semantic rule interpretation:
(1) matching rules:
1) all represents all sub-elements in the phrase that must be matched;
2) anyone which represents any one of the neutron elements in the matching phrase;
3) none, meaning that any one of the sub-elements in the phrase cannot be matched;
4) the notall indicates that all the sub-elements in the phrase are not matched;
(2) matching algorithm:
1) comprising (continain): the natural language input by the user comprises specified elements;
2) similarity (like): calculating the similarity of the natural language and the designated elements input by the user by using a similarity algorithm until a designated threshold value is reached;
3) regular expression (reg): matching the natural language input by the user with a specified element (a regular matching pattern);
4) custom matching algorithms can also be extended and added as needed.
II, introduction:
1. determining a plurality of expression modes of the semantics, namely defining SREA;
2. defining a plurality of semantic sub-rules for each expression mode, namely defining SSR;
3. the logical relationship among a plurality of sub-rules is expressed by a semantic rule expression, namely, an SRE is defined;
4. then, making detailed definition for each specific semantic sub-rule, and determining a specific matching algorithm, a matching rule and a matching phrase to be used, namely definition f (MR, MWA, MA);
after the above work is completed, the definition of a specific semantic matching implementation is completed.
Thirdly, semantic matching-logic flow:
1. acquiring a Natural Language (NL) input by a user and a semantic meaning (S) to be matched;
2. searching a Semantic Rule Expression Array (SREA) of the semantic;
3. traversing each Semantic Regular Expression (SRE) in the SREA;
4. and judging whether the SRE is successfully matched (referring to an SRE matching process), wherein the semantic meaning is successfully matched as long as one SRE is successfully matched.
Fourthly, Semantic Rule Expression (SRE) matching-logic flow:
1. the semantic rule expression comprises a plurality of semantic sub-rules (SSRs), and logical operation relations exist among the SSRs;
2. and judging whether each SSR is successfully matched (referring to an SSR matching process), calculating the matching results of a plurality of SSRs by using a logic calculator, and judging whether the whole SRE is successfully matched.
Fifthly, semantic sub-rule (SSR) matching-logic flow:
1. a specific SSR match, which can be represented by a function f (MR, MWA, MA);
2. the Natural Language (NL) input by the user needs to use the function to judge whether the semantic sub-rule can be successfully matched with the natural language;
3. different Matching Rules (MR) need to use different matching logics, and whether the SSR is successfully matched is judged according to the rule matching result;
4. matching rules:
(1) for all sub-elements in a matching phrase (MWA), a Matching Algorithm (MA) is used to determine whether the Natural Language (NL) entered by the user matches the sub-element:
1) when the matching rule is all, only if all the sub-elements are successfully matched, the matching rule calculates that the matching is successful;
2) when the matching rule is anyone, as long as any sub-element is successfully matched, the matching rule is successfully matched;
3) when the matching rule is none, all the sub-elements cannot be successfully matched, and the matching rule is calculated to be successfully matched;
4) when the matching rule is notall, as long as all the sub-elements are not successfully matched, the matching rule is successful even if the matching is successful;
(2) matching algorithm:
1) when the matching algorithm is contained (container), the matching algorithm succeeds even if NL contains the child element;
2) when the matching algorithm is similarity (like), calculating the similarity between NL and a sub-element (the scene is generally a sentence) by using a text similarity algorithm, and if the similarity exceeds a specified threshold, the matching algorithm is successful;
3) when the matching algorithm is a regular match (reg), using a regular match NL and a sub-element (such a scene is generally a matching pattern), as long as the NL and the sub-element match successfully, the matching algorithm is successful even if the matching is successful;
4) when the matching algorithm is other self-defined matching algorithms, other matching algorithm functions or self-realized matching algorithms can be called to judge whether the matching algorithm is successfully matched.
Example (c): assume that the matching rules and matching algorithms are denoted with the following identifiers:
1. matching rules:
R1=all
R2=anyone
R3=none
R4=notall
2. and (3) matching algorithm:
A1=contain
a2 (threshold) = like
A3=reg
The semantic "user expresses a desire to see a movie" can be defined by:
semantic = user expressing wanting to watch a movie
Semantic rule expression 1 = R3a1[ not ] & R2a1[ think | like ] & R2a1[ see ] & R2a1[ movie | piece ]
Semantic rule expression 2 = R2a1[ long | very recent ] & R2a1[ none ] & R2a1[ cinema | watch movie ]
Semantic rule expression 3 = R3a1[ not ] & R2a2(0.8) [ want to see movie | do not see movie for a long time ]
Semantic rule expression 4 = R3a1[ not ] & R2A3[. this is (thinking of | like). this is (movie | piece) ]
The invention realizes the whole semantic matching through a text similarity algorithm, a logic calculator algorithm and a regular matching algorithm.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (5)
1. A general semantic matching implementation method is characterized by comprising the following steps:
s1, acquiring the natural language input by the user and the semantic to be matched;
s2, semantic matching, comprising the following steps:
s21, searching a semantic rule expression array SREA corresponding to the semantics;
s22, traversing each semantic regular expression SRE in the semantic regular expression array SREA;
s23, judging whether the natural language input by the user is matched with the semantic rule expression SRE, and judging whether the natural language input by the user is successfully matched with each semantic sub-rule SSR or not, wherein the matching results of the plurality of semantic sub-rule SSRs are calculated by a logic calculator, so that whether the whole semantic rule expression SRE is successfully matched or not is judged, if only one SRE is successfully matched, the natural language input by the user is successfully matched with the semantic, and if not, the matching is failed; wherein the matching of the semantic sub-rule SSR to the natural language input by the user is represented by a function f (MR, MWA, MA), comprising the steps of:
s231, for all matched sub-elements MSE in the matched phrase MWA, a matching algorithm MA is used for judging whether the natural language input by the user is matched with the matched sub-elements MSE:
when the matching rule MR is all, the matching rule MR calculates that the matching is successful only if all the matching sub-elements MSE are successfully matched;
when the matching rule MR is anyone, as long as any matching sub-element MSE is successfully matched, the matching rule MR is even successfully matched;
when the matching rule MR is none, all the matching sub-elements MSE can not be successfully matched, and the matching rule MR is only calculated to be successfully matched;
when the matching rule MR is notall, as long as all the matching sub-elements MSE are not successfully matched, the matching rule MR is even if the matching is successful;
s232, matching algorithm MA:
when the matching algorithm MA is contained, as long as the natural language input by the user contains the matching sub-element MSE, the matching algorithm MA is successful even if the matching is successful;
when the matching algorithm MA is similarity, calculating the similarity between the natural language input by the user and the matching sub-element MSE, and if the similarity exceeds a specified threshold, successfully matching the matching algorithm MA;
when the matching algorithm MA is a regular expression, the regular expression is used for matching the natural language input by the user and the matched sub-element MSE, and the matching algorithm MA is even matched successfully as long as the natural language input by the user and the matched sub-element MSE are matched successfully;
when the matching algorithm is other self-defined matching algorithms, other matching algorithm functions or self-realized matching algorithms can be called to judge whether the matching algorithm is successfully matched.
2. The method as claimed in claim 1, wherein in step S232, when the matching algorithm MA is similarity, a text similarity algorithm is used to calculate similarity between the natural language input by the user and the sentence as the matching sub-element MSE.
3. The method as claimed in claim 1, wherein in step S232, when the matching algorithm MA is a regular expression, the regular expression is used to match the natural language input by the user and the matching pattern as the matching sub-element MSE.
4. A general semantic matching implementation system comprises a semantic rule expression array SREA, and is characterized in that the semantic rule expression array SREA comprises a plurality of semantic rule expressions SRE which comprise a plurality of semantic sub-rules SSR connected through logic relations, one semantic has a plurality of expression modes, whether a natural language input by a user is matched with the semantic rule expression SRE is judged by traversing each semantic rule expression SRE in the semantic rule expression array SREA, if one SRE is successfully matched, the natural language input by the user is successfully matched with the semantic rule expression SRE, otherwise, the matching is failed;
the semantic sub-rule SSR comprises a matching rule MR, a matching phrase MWA and a matching algorithm MA; the matching rule MR is used to match the rule of the phrase MWA, including all: representing that all matched subelements MSE in the matched phrase MWA must be matched; and (3) any one of: any matching sub-element MSE in the matching phrase MWA is shown to be matched; none: indicating that it cannot match any of the matching sub-elements MSE in the matching phrase MWA; nodall: the expression is only required to match all the matching sub-elements MSE in the matched phrase MWA; the matching phrase MWA includes: matching sub-elements MSE, under different configuration algorithms MA, having different types of matching sub-elements MSE; the matching algorithm MA is used for matching the natural language input by the user with the matching sub-element MSE in the matching phrase MWA, and comprises the following steps: comprises the following steps: the natural language input by the user comprises the specified matching sub-element MSE; similarity: calculating the similarity of the natural language input by the user and the specified matching sub-element MSE by using a similarity calculation method until the similarity reaches a specified threshold; the regular expression is as follows: matching the natural language input by the user with the specified matching sub-element MSE; self-defining: self-expanding and adding according to the requirement; the matching sub-element MSE is a word, a sentence or a regular matching pattern.
5. A generic semantic matching implementation system according to claim 4, characterized in that the logical relations comprise AND, OR, NOT, brackets.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110179941.6A CN112732876B (en) | 2021-02-08 | 2021-02-08 | Universal semantic matching implementation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110179941.6A CN112732876B (en) | 2021-02-08 | 2021-02-08 | Universal semantic matching implementation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112732876A CN112732876A (en) | 2021-04-30 |
| CN112732876B true CN112732876B (en) | 2022-07-08 |
Family
ID=75596639
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110179941.6A Active CN112732876B (en) | 2021-02-08 | 2021-02-08 | Universal semantic matching implementation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112732876B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115577699B (en) * | 2022-12-09 | 2023-04-14 | 杭州北冥星眸科技有限公司 | Method for determining text entry reasonableness, electronic equipment and storage medium |
| CN116757203B (en) * | 2023-08-16 | 2023-11-10 | 杭州北冥星火科技有限公司 | Natural language matching method, device, computer equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104199803A (en) * | 2014-07-21 | 2014-12-10 | 安徽华贞信息科技有限公司 | Text information processing system and method based on combinational theory |
| CN108280225A (en) * | 2018-02-12 | 2018-07-13 | 北京吉高软件有限公司 | A kind of semantic retrieving method and searching system |
| CN111191105A (en) * | 2019-10-31 | 2020-05-22 | 腾讯科技（深圳）有限公司 | Method, device, system, equipment and storage medium for searching government affair information |
| CN111324708A (en) * | 2020-02-07 | 2020-06-23 | 普强时代(珠海横琴)信息技术有限公司 | Natural language processing system based on human-computer interaction |
| CN111898386A (en) * | 2020-07-23 | 2020-11-06 | 上海京知信息科技有限公司 | Semantic parsing method, device, equipment and storage medium based on natural language |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9898529B2 (en) * | 2014-06-30 | 2018-02-20 | International Business Machines Corporation | Augmenting semantic models based on morphological rules |
-
2021
- 2021-02-08 CN CN202110179941.6A patent/CN112732876B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104199803A (en) * | 2014-07-21 | 2014-12-10 | 安徽华贞信息科技有限公司 | Text information processing system and method based on combinational theory |
| CN108280225A (en) * | 2018-02-12 | 2018-07-13 | 北京吉高软件有限公司 | A kind of semantic retrieving method and searching system |
| CN111191105A (en) * | 2019-10-31 | 2020-05-22 | 腾讯科技（深圳）有限公司 | Method, device, system, equipment and storage medium for searching government affair information |
| CN111324708A (en) * | 2020-02-07 | 2020-06-23 | 普强时代(珠海横琴)信息技术有限公司 | Natural language processing system based on human-computer interaction |
| CN111898386A (en) * | 2020-07-23 | 2020-11-06 | 上海京知信息科技有限公司 | Semantic parsing method, device, equipment and storage medium based on natural language |
Non-Patent Citations (2)
| Title |
|---|
| Offset-FA: Detach the Closures and Countings for Efficient Regular Expression Matching;Chengcheng Xu et al.;《2017 IEEE 7th International Symposium on Cloud and Service Computing (SC2)》;20171125;第263-266页 * |
| 类型-逻辑语法——一种自然语言分析的重要工具;张秋成;《中山大学学报（社会科学版）》;20031230;第57-66页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112732876A (en) | 2021-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7223785B2 (en) | TIME-SERIES KNOWLEDGE GRAPH GENERATION METHOD, APPARATUS, DEVICE AND MEDIUM | |
| Chakraborty et al. | Introduction to neural network based approaches for question answering over knowledge graphs | |
| CN112732876B (en) | Universal semantic matching implementation method and system | |
| Kwiatkowski et al. | Scaling semantic parsers with on-the-fly ontology matching | |
| US8762130B1 (en) | Systems and methods for natural language processing including morphological analysis, lemmatizing, spell checking and grammar checking | |
| US20130204611A1 (en) | Textual entailment recognition apparatus, textual entailment recognition method, and computer-readable recording medium | |
| Chakraborty et al. | Introduction to neural network‐based question answering over knowledge graphs | |
| US20230259707A1 (en) | Systems and methods for natural language processing (nlp) model robustness determination | |
| Coavoux et al. | Neural greedy constituent parsing with dynamic oracles | |
| Zhu et al. | MDCSpell: A multi-task detector-corrector framework for Chinese spelling correction | |
| KR20220127786A (en) | Data processing method and apparatus for training neural networks classifing natural language intentions | |
| Fan et al. | Uncovering the potential of chatgpt for discourse analysis in dialogue: An empirical study | |
| Huang et al. | Chinese spelling check system based on tri-gram model | |
| Luo et al. | Logic against bias: Textual entailment mitigates stereotypical sentence reasoning | |
| Van der Goot | Machamp at semeval-2022 tasks 2, 3, 4, 6, 10, 11, and 12: Multi-task multi-lingual learning for a pre-selected set of semantic datasets | |
| Wintner | Formal language theory | |
| Yang et al. | Spell Checking for Chinese. | |
| Angluin et al. | A model of semantics and corrections in language learning | |
| Schwind | Sensitive parsing: error analysis and explanation in an intelligent language tutoring system | |
| Lo et al. | GPoeT-2: A GPT-2 based poem generator | |
| Pu et al. | Passing parser uncertainty to the transformer: Labeled dependency distributions for neural machine translation | |
| JP6273285B2 (en) | Format module, system and method for formatting electronic string | |
| Vrublevskyi et al. | Grammar Error Correcting by the Means of CFG Parser | |
| CN115688748A (en) | Question error correction method and device, electronic equipment and storage medium | |
| Sheng et al. | EDMSpell: Incorporating the error discriminator mechanism into chinese spelling correction for the overcorrection problem |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| PE01 | Entry into force of the registration of the contract for pledge of patent right |
Denomination of invention: A Universal Semantic Matching Implementation Method and System Granted publication date: 20220708 Pledgee: Zhejiang Mintai Commercial Bank Co.,Ltd. Hangzhou Binjiang small and micro enterprise franchise sub branch Pledgor: HANGZHOU GALAXY EYE TECHNOLOGY Co.,Ltd. Registration number: Y2024330000566 |