CN111625640B - Question and answer processing method, device and storage medium - Google Patents
Question and answer processing method, device and storage medium Download PDFInfo
- Publication number
- CN111625640B CN111625640B CN202010529247.8A CN202010529247A CN111625640B CN 111625640 B CN111625640 B CN 111625640B CN 202010529247 A CN202010529247 A CN 202010529247A CN 111625640 B CN111625640 B CN 111625640B
- Authority
- CN
- China
- Prior art keywords
- entity
- target
- pair
- relation
- question
- 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
- 238000003672 processing method Methods 0.000 title claims abstract description 23
- 238000000034 method Methods 0.000 claims abstract description 47
- 238000005065 mining Methods 0.000 claims abstract description 33
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000013507 mapping Methods 0.000 claims abstract description 22
- 238000001914 filtration Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 238000010586 diagram Methods 0.000 description 5
- 238000010276 construction Methods 0.000 description 4
- 238000002360 preparation method Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 229920000433 Lyocell Polymers 0.000 description 1
- 108091081062 Repeated sequence (DNA) Proteins 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
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
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- 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
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a question and answer processing method, a question and answer processing device and a storage medium. The method comprises the steps of mining entity relation pairs based on a corpus, mining and forming a problem cluster mode based on a problem, establishing a mapping relation between the problem cluster mode and a problem cluster type, determining a target entity relation pair and a target problem cluster mode corresponding to the problem by combining the entity relation pair and the problem cluster mode for a given problem, determining a target problem cluster type corresponding to the target problem cluster mode based on the mapping relation between the problem cluster mode and the problem cluster type, and generating an answer to the problem according to the relation pair entity in the target entity relation pair when the entity relation between the target problem cluster type and the target entity relation pair forms a preset relation.
Description
Technical Field
The present invention relates to the field of information retrieval and processing technologies, and in particular, to a question-answering processing method, device and storage medium.
Background
Several types of automated question-answering systems are common: from question to question, based on structured knowledge base, based on unstructured data, chat bots. At present, several major methods for constructing question-answering systems mainly comprise: question-answering systems based on similarity of questions, search-based question-answering systems, structured data-based question-answering systems.
Firstly, a question-answering system based on the similarity of questions must be established, wherein a common question answer library with a certain scale is firstly established, the common question answer library stores a large number of questions and answers corresponding to the questions, when a user puts forward a new question, sentence similarity calculation is carried out on the proposed question and all questions in the question answer library, when the similarity exceeds a set threshold value, answer recall corresponding to the question with the highest similarity value in the question answer library is fed back, and if the similarity value does not exceed the set threshold value, the system selects to answer the question without answer;
firstly, a question and answer system based on search needs to identify the intention of a question, and the identification method comprises the following steps: searching based on the query words and the centering relation, finding possible answer types, when a plurality of answer types occur, often finding a word with high generalization degree based on word frequency, generalization degree of the words, based on dependency syntactic relation and the like, or using the word with the previous problem as the answer type, then marking the answer to be selected and extracting the characteristics, wherein the answer to be selected is obtained by marking and extracting the search result of a search engine, and then scoring and sorting each answer, constructing an answer sorting model, and feeding back the answer with the highest recall score;
The question-answering system based on the structured data adopts pattern matching, such as the structured data based on the knowledge graph, firstly carries out grammar and semantic analysis on the questions, then inquires related information in a given knowledge base, returns the most related information as answers to the questions according to the degree of relativity and feeds back the answers to the questions to the user.
The above-mentioned method for constructing a question-answering system at present has problems in that:
1. the question-answering system based on the similarity of the questions needs to construct a question answer library with a larger scale, the similarity between the question answer library and sentences of each question in the question answer library needs to be calculated for given questions, the calculated amount is larger, and the time consumption on line is longer;
2. the question-answering system based on search depends on the search efficiency and result of a search engine, and simultaneously needs to combine the syntactic relation and answer labels to perform feature extraction, which is influenced by the extraction accuracy of the syntactic relation and the coverage rate of the defined rules;
3. the question-answering system based on the structured data needs to construct a knowledge graph first, then a series of modes are formulated for matching, and more earlier work is needed.
Disclosure of Invention
The invention provides a question and answer processing method, a question and answer processing device and a storage medium, which can rapidly output accurate answers to given questions.
In a first aspect, the present invention provides a question-answering processing method, including:
acquiring a target problem;
performing entity relation pair matching on the target problem according to an entity relation pair set, and determining a target entity relation pair corresponding to the target problem, wherein the entity relation pair set comprises at least one entity relation pair, and each entity relation pair comprises two relation pair entities and an entity relation between the two relation pair entities;
performing problem cluster pattern matching on the target problem based on a problem cluster pattern set, and determining a target problem cluster pattern corresponding to the target problem, wherein the problem cluster pattern set comprises at least one problem cluster pattern;
determining a target problem cluster type corresponding to the target problem cluster mode according to the mapping relation between the problem cluster mode and the problem cluster type;
when a preset relationship is formed between the target problem cluster type and the entity relationship in the target entity relationship pair, generating an answer to the target problem according to the relationship in the target entity relationship pair.
In a second aspect, the present invention provides a question-answering processing apparatus, including:
the problem acquisition module is used for acquiring target problems;
The entity relation pair matching module is used for carrying out entity relation pair matching on the target problem according to an entity relation pair set, determining a target entity relation pair corresponding to the target problem, wherein the entity relation pair set comprises at least one entity relation pair, and each entity relation pair comprises two relation pair entities and entity relations between the two relation pair entities;
the problem cluster pattern matching module is used for carrying out problem cluster pattern matching on the target problem based on a problem cluster pattern set, determining a target problem cluster pattern corresponding to the target problem, wherein the problem cluster pattern set comprises at least one problem cluster pattern obtained by carrying out frequent sequence pattern mining on a problem set sample;
the problem cluster type determining module is used for determining a target problem cluster type corresponding to the target problem cluster mode according to the mapping relation between the problem cluster mode and the problem cluster type;
and the answer determining module is used for generating an answer to the target question for the entity according to the relation pair in the target entity relation pair when a preset relation is formed between the target question cluster type and the entity relation in the target entity relation pair.
In a third aspect, the present invention provides a computer-readable storage medium having stored therein at least one instruction or at least one program loaded and executed by a processor to implement the question-answering processing method as provided in the first aspect.
The question-answering processing method, the question-answering processing device and the storage medium provided by the invention have the following technical effects:
according to the method, the entity relation pairs are mined based on the corpus, the problem cluster patterns are formed based on problem mining, the mapping relation between the problem cluster patterns and the problem cluster types is established, for a given problem, the target entity relation pairs and the target problem cluster patterns corresponding to the problem are determined by combining the entity relation pairs and the problem cluster patterns, then the target problem cluster types corresponding to the target problem cluster patterns are determined based on the mapping relation between the problem cluster patterns and the problem cluster types, when the entity relation between the target problem cluster types and the target entity relation pairs forms a preset relation, answers to the problem are generated for the entity pairs according to the relation in the target entity relation pairs, and the scheme improves the accuracy of the answers and the answer recall efficiency by fusing the problem cluster patterns and the entity relation pairs.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic diagram of an optional application scenario of a question-answering processing method provided by an embodiment of the present invention;
FIG. 2 is a data sharing system shown in an embodiment of the present invention;
fig. 3 is a schematic flow chart of a question-answering processing method according to an embodiment of the present invention;
FIG. 4 is a flow chart of a method for constructing a problem cluster pattern set according to an embodiment of the present invention;
FIG. 5 is a flow chart of a method for building a set of entity relationship pairs provided by an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a question-answering processing device according to an embodiment of the present invention;
fig. 7 is a schematic structural diagram of an embodiment of a server provided by the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or server that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Fig. 1 is a schematic diagram of an optional application scenario of the question-answering processing method provided in the embodiment of the present invention, please refer to fig. 1, in which a terminal 200 is connected to a server 100 through a network 300, and the network 300 may be a wide area network or a local area network, or a combination of the two.
The terminal 200 may be used to acquire problem information, for example, when a user inputs problem information through an input interface, the terminal automatically acquires problem information after the input is completed.
In some embodiments, the terminal 200 may send, to the server 100 through the network 300, question information input by a user on the terminal 200, and invoke a question-answer processing function provided by the server 100, where the server 100 obtains an accurate answer through the question-answer processing method provided by the embodiment of the present invention, for example, a search APP is installed on the terminal 200, the user inputs the question information in the search APP, the terminal 200 sends the question information to the server 100 through the network 300, the server 100 performs a series of mining processes according to the question information, obtains an accurate answer, and returns the answer to the search APP, and the answer is displayed on a display interface of the terminal 200.
The terminal 200 may include: the smart phone, tablet computer, notebook computer, digital assistant, intelligent wearable device, vehicle terminal and other entity devices can also include software running in the entity devices, such as application programs with question and answer processing functions. The terminal 200 may be communicatively connected to the Server 100 based on Browser/Server (B/S) or Client/Server (C/S) mode.
The server 100 may comprise a single independently operating server, or a distributed server, or a server cluster consisting of a plurality of servers.
According to the conception of the invention, a server excavates entity relation pairs based on a corpus in advance, performs repeated sequence pattern excavation on questions and samples to obtain a question cluster pattern set, establishes mapping relations between each question cluster pattern and question cluster types in the question cluster pattern set, and performs entity relation pair matching on target questions according to the entity relation pair set when receiving questions sent by a terminal to determine target entity relation pairs corresponding to the target questions; performing problem cluster pattern matching on the target problem based on the problem cluster pattern set, and determining a target problem cluster pattern corresponding to the target problem; determining a target problem cluster type corresponding to the target problem cluster mode according to the mapping relation between the problem cluster mode and the problem cluster type; when a preset relation is formed between the target problem cluster type and the entity relation in the target entity relation pair, generating an answer to the target problem according to the relation in the target entity relation pair. Therefore, the answer type identification is not needed for each question, but when the questions meet the preset relationship between the entity and the relationship pair entity, and the entity relationship of the entity relationship pair and the type of the question cluster, the answer entity can be accurately matched, the question and answer can be realized by combining the text mode and the entity relationship, the answer recall target is more definite, and the recall efficiency is higher. A detailed description about this scheme will be described later with reference to fig. 3 to 6.
The server in the scenario of the question-answering processing method according to the embodiment of the present invention may be a data sharing system formed by connecting a plurality of nodes (any form of computing devices in an access network, such as servers, clients) through a network communication.
Referring to the data sharing system shown in fig. 2, the data sharing system 400 refers to a system for performing data sharing between nodes, and the data sharing system may include a plurality of nodes 101, and the plurality of nodes 101 may be respective clients in the data sharing system. Each node 101 may receive input information while operating normally and maintain shared data within the data sharing system based on the received input information. In order to ensure the information intercommunication in the data sharing system, information connection can exist between each node in the data sharing system, and the nodes can transmit information through the information connection. For example, when any node in the data sharing system receives input information, other nodes in the data sharing system acquire the input information according to a consensus algorithm, and store the input information as data in the shared data, so that the data stored on all nodes in the data sharing system are consistent.
Each node in the data sharing system has a node identifier corresponding to the node identifier, and each node in the data sharing system can store the node identifiers of other nodes in the data sharing system, so that the generated block can be broadcast to other nodes in the data sharing system according to the node identifiers of other nodes. Each node can maintain a node identification list shown in the following table, and the node names and the node identifications are correspondingly stored in the node identification list. The node identifier may be an IP (Internet Protocol, protocol interconnecting between networks) address, or any other information that can be used to identify the node.
An embodiment of the question-answering method according to the present invention is described below, and fig. 3 is a schematic flow chart of the question-answering method according to the embodiment of the present invention, and the present specification provides the steps of the method according to the embodiment or the flowchart, but may include more or less steps based on conventional or non-inventive labor. The order of steps recited in the embodiments is merely one way of performing the order of steps and does not represent a unique order of execution. In actual system products, the processes may execute sequentially or in parallel (e.g., in a parallel processor or a multithreaded environment) in accordance with the methods shown in the embodiments or figures. As shown in fig. 3, the question-answering processing method is executed by the server, and may include a data preparation stage and a question-answering processing stage, where a user determines a set of entity relation pairs, a set of question cluster patterns, and a mapping relation between each question cluster pattern and a question cluster type in the set of question cluster patterns, and the question-answering processing stage generates an answer to a question based on the above data determined in the data preparation stage.
Data preparation stage
The method comprises the steps of constructing an entity relation pair set, a problem cluster mode set and a mapping relation between a problem cluster mode and a problem cluster type in the problem cluster mode set.
The method for constructing the entity relation pair set comprises the following steps:
first, a corpus is obtained, wherein the corpus comprises at least one corpus.
The corpus in the corpus set may be information crawled from a web page, for example, extracted from hundred-degree vocabulary entries, news information, wikipedia.
And secondly, analyzing and processing the corpus to determine an entity relation pair set corresponding to the corpus.
FIG. 5 is a flow chart of a method for building a set of entity relationship pairs according to an embodiment of the present invention. Referring to fig. 5, for each corpus, its corresponding set of entity-relationship pairs may be determined by:
s501, acquiring entities contained in the corpus to obtain an entity set, wherein the entity set comprises at least two entities.
The entity refers to entity words in the corpus, including nouns and pronouns, the corpus can be preprocessed before determining the entity contained in the corpus, the pronouns are replaced by corresponding nouns, and then nouns contained in the corpus are extracted as the entity.
S503, combining the entities in the entity set in pairs to form at least one entity pair, wherein each entity pair comprises two entities.
S505, mining the relation between two entities in each entity pair according to the corpus to obtain the entity relation corresponding to each entity pair, and forming the entity relation pair according to the entity pair and the entity relation corresponding to the entity pair.
In one possible embodiment, the following can be done for each entity pair:
counting word frequencies of words in the context where two entities in the entity relation pair are located together; when word frequency of a word exceeds a set threshold, taking the two entities as relationship pair entities, and taking the word as an entity relationship between the two relationship pair entities; and establishing a mapping relation between the entity relation and the two relation pair entities to form an entity relation pair.
S507, constructing an entity relation pair set based on the composed entity relation pairs.
In a specific application, entity relation pairs can be constructed based on webpage data such as hundred-degree terms, news information, wikipedia and the like to mine the relation between the entities, and the obtained entity relation pairs are stored in a triplet corpus of the entity relation pairs for subsequent retrieval. An example of a pair of entity relationships to mining results is shown in table (1).
| Entity 1 | Entity relationship | Entity 2 |
| Thousand sail live broadcast | Company (Corp) | Fox searching device |
| Tencel video | Year of year | 2011 |
| Green XX | Composition of music | XXX |
Watch (1)
Fig. 4 is a flowchart of a method for constructing a problem cluster pattern set according to an embodiment of the present invention, please refer to fig. 4, wherein the method for constructing a problem cluster pattern set includes:
s401, acquiring a question set sample, wherein the question set sample comprises at least one question.
S403, mining frequent word sequence patterns of each length meeting the minimum support requirement in the problem set sample by taking the word sequence as a frequent sequence pattern mining object, and obtaining a first problem pattern set, wherein the first problem pattern set comprises at least two problem sequence patterns.
In one possible embodiment, step S403 may include mining frequent word sequence patterns of lengths in the question set sample that meet a minimum support requirement:
splitting word elements and word elements of each sentence in the problem set sample to obtain first sample data; determining the support degree of each word element and each word element according to the occurrence times of the word element and the word element in sentences contained in the first sample data; removing word elements and word elements with the support degree smaller than a preset support degree threshold value in the first sample data to obtain second sample data; taking the word elements and the word elements with the support degree not smaller than a preset support degree threshold value as prefixes to obtain a prefix set; determining a projection data set which corresponds to each prefix and consists of word elements and/or word elements which follow the prefix according to the occurrence condition of each prefix in the prefix set in the second sample data; and performing recursive mining operation on each projection data set to obtain all target prefixes meeting the minimum support requirement, wherein each target prefix is a problem sequence pattern obtained by mining a problem set sample, and generating a first problem pattern set according to all target prefixes obtained by mining.
Specifically, performing a recursive mining operation on each of the projection data sets may include:
judging whether the projection data set is an empty set or not; recursively returning if the projection data set is an empty set; if the projection data set is not an empty set, counting the support degree of each word element and each word element in the projection data set, and judging whether the support degree of each word element and each word element meets the minimum support degree requirement; if the support degree of the character element or the word element does not meet the minimum support degree requirement, recursively returning; if the support degree of the character element or the word element meets the minimum support degree requirement, merging the prefixes corresponding to the character element or the word element and the projection data set to obtain a new prefix, determining the projection data set of the new prefix, and performing recursion mining on the projection data set corresponding to the new prefix to obtain all target prefixes meeting the minimum support degree requirement; the recursively returned data are all target prefixes which are currently obtained and meet the minimum support requirement and the corresponding support.
S405, judging the inclusion relation of the problem sequence modes in the first problem mode set, and determining the category of each problem sequence mode, wherein the category comprises a subsequence and a supersequence.
Specifically, when judging the inclusion relation of the problem sequence pattern, determining a corresponding item set according to words contained in the problem sequence pattern, for example, the item set of the problem sequence pattern "which year to release" is { yes, which year, release, product }, "yes", "which year", "release", and "product" are items of the item set; if all items in the set of items of problem sequence pattern A are found in the set of items in another problem sequence pattern B, then A is a subsequence of B and B is a supersequence of A.
S407, removing the problem sequence modes with the sub-sequences in the first problem mode set to obtain a second problem mode set consisting of the problem sequence modes with the super-sequences.
S409, filtering each problem sequence mode in the second problem mode set, and taking the problem sequence mode obtained by the filtering as a problem cluster mode.
In a possible embodiment, a specific means of filtering the question sequence pattern may be to filter out the entity words appearing in the question sequence pattern, for example, the question sequence pattern "who is the blue XX", where "blue XX" is the entity word, and the question sequence pattern obtained after the filtering is "who is the blue XX".
S411, constructing a problem cluster mode set according to the filtered problem cluster modes.
The inventor finds out through data analysis that for thousands of question-answer pairs, mode information is often hidden in the questions, and the mode information represents common answer types (such as time, characters, places and the like) of the questions.
In specific implementation, frequent word sequence patterns of each length meeting the minimum support degree condition in the problem context can be mined based on a Prefixspan algorithm, and the condition meeting the minimum support degree refers to that the support degree is not smaller than a minimum support degree threshold value. The calculation method of the minimum support is shown in formula (1).
min_sup＝a×n (1)
Wherein n is the number of samples of the mining problem set, a is the minimum support rate, and the minimum support rate parameter is adjusted according to the number of samples of the problem set.
The specific operation steps of the Prefixspan algorithm are as follows:
1. finding out a prefix of a problem word sequence with unit length of 1 and a corresponding projection data set;
2. counting the occurrence frequency of the prefix of the word sequence of the problem, adding the prefix with the support degree higher than the minimum support degree threshold value into the data set, and obtaining a frequent one-set sequence mode;
3. Recursively mining all prefixes of length i and meeting minimum support requirements:
1) Mining the projection data set of the prefix, and returning to recursion if the projection data is an empty set;
2) Counting the minimum support degree of each item in the corresponding projection data set, combining each item meeting the support degree with the current prefix to obtain a new prefix, and recursively returning if the support degree requirement is not met;
3) Let i=i+1, the prefixes are each new prefix after merging the single items, and the 3 rd step is executed recursively respectively;
4. and returning all frequent sequence modes in the problem set sample, wherein the frequent sequence modes are the problem sequence modes.
The above is a principle description of the Prefixspan algorithm and the following examples illustrate specific mining processes.
Data samples for the example problem set samples are shown in table (2):
watch (2)
Based on the Prefixspan algorithm, mining sequence patterns contained in the problem set samples, assuming that the set minimum support rate threshold value is 0.5, firstly counting the number of occurrence samples of all word sequence elements, then filtering word sequence elements which do not accord with the preset support rate threshold value, wherein the set minimum support rate threshold value is 0.2, namely that at least 2 samples of word elements in the above 6 samples can not be smaller than the minimum support rate threshold value, and the statistical result of the word elements which are not smaller than the minimum support rate threshold value is shown in a table (3):
| Word and word | A kind of electronic device | Is that | Green XX | Product(s) | Publishing | Company (Corp) | Which one | Which is | Year of life | Who is |
| Number of appearance samples | 6 | 6 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 |
Watch (3)
After filtering out word elements less than the minimum support threshold, the pre-processed samples are shown in table (4):
| which company's product |
| Which is of the company |
| Is issued in which year |
| Is a product released in which year |
| Whose green XX is |
| Whose green XX is |
Watch (4)
Constructing a prefix and a corresponding suffix of the word sequence element meeting the threshold, wherein the result is shown in a table (5):
watch (5)
Taking the two-term prefix as a 'yes' example, continuing to mine the two-term prefix and the corresponding suffix meeting the minimum support degree condition, wherein the two-term prefix and the corresponding suffix are shown in a table (6):
watch (6)
Taking two prefixes as 'which is, who is' as an example, three prefixes and corresponding suffixes meeting the minimum support condition are continuously mined, and the three prefixes and the corresponding suffixes are shown in a table (7):
watch (7)
Continuing to mine four prefixes and corresponding suffixes meeting the minimum support condition, wherein the four prefixes and the corresponding suffixes are shown in a table (8):
watch (8)
Continuing to mine the five-item prefix and the corresponding suffix meeting the minimum support degree condition, wherein the five-item prefix and the corresponding suffix are shown in a table (9):
| five-item prefix | Corresponding suffix |
| Is issued in which year | Product(s) |
Watch (9)
After iteration is finished, cleaning the mined problem sequence mode, and specifically, the steps comprise:
1. in the same sample, the inclusion relation judgment is carried out on the problem sequence patterns of each length of the mining, and the sub-sequence pattern filtering is carried out.
In the same sample, if all items of the item set of a certain problem sequence pattern a can be found in the item set of the problem sequence pattern B, a is a subsequence of B, and B is a supersequence of a. According to this definition, for the item set a= { a of problem sequence pattern a 1 ,a 2 ,...a n Item set b= { B of } and problem sequence pattern B 1 ,b 2 ,...b n M, n.ltoreq.m, if the number sequence 1.ltoreq.j is present 1 ≤j 2 ≤...≤j n M is less than or equal to m, meetsThen a is called a subsequence of B. For each problem sequence mode obtained by mining, if the supersequence itself contains more reference information, namely the supersequence contains a context auxiliary word which is not contained in the subsequence, the supersequence is reserved as a mode, and the subsequence is deleted.
In the case of the same sample "what year the Tengbing video is released", for example, the supersequence "which year the supersequence is released" and the subsequence "which year the supersequence is released" are taken, and the supersequence further includes the context auxiliary information "year, and therefore, the supersequence mode is reserved and the subsequence mode is deleted on the basis of the subsequence.
The entity words that appear in the pattern are then filtered, including the specific thing, individual subjects, etc., as in the example above, "cyan XX". The final mining results in a problem cluster pattern as shown in table (10):
| which company's product |
| Which is of the company |
| Is issued in which year |
| Is a product released in which year |
| Who is of |
Watch (10)
Then, constructing a mapping relation between the problem cluster mode and the problem cluster type in the problem cluster mode set, wherein the mapping relation comprises the following steps: determining a problem cluster type of each problem cluster mode in the problem cluster mode set; and establishing a mapping relation between the problem cluster type and the problem cluster mode.
Based on the question cluster pattern obtained by the mining in the steps, marking the question cluster type word of the question cluster pattern, wherein the question cluster type word is an answer type word, so that the answer type of the question cluster pattern is positioned, and the question cluster type word is as follows: time, place, person, event, etc. may also be labeled in more detail.
For example, for the above problem cluster patterns, the type of the problem cluster corresponding to each pattern is shown in table (11):
watch (11)
(II) question-answer processing stage
The question-answer processing stage is to match answers corresponding to the questions by using the data acquired in the data preparation stage. Fig. 3 is a flow chart of a question-answer processing method provided by an embodiment of the present invention, please refer to fig. 3, the question-answer processing method includes:
s301: and acquiring a target problem.
S303: and carrying out entity relation pair matching on the target problem according to an entity relation pair set, and determining a target entity relation pair corresponding to the target problem, wherein the entity relation pair set comprises at least one entity relation pair, and each entity relation pair comprises two relation pair entities and an entity relation between the two relation pair entities.
In one possible embodiment, the step includes:
performing entity matching on the target problem by the entity according to the relation contained in the entity relation pair set to obtain a problem entity corresponding to the target problem; and determining a target entity relation pair corresponding to the problem entity according to the subordinate relation between each relation pair entity and the entity relation pair in the entity relation pair set.
The target question is presented in question form, including subject, predicate, and object. In some embodiments, if a subject of a target problem matches a relationship pair entity contained in a set of relationship pairs of entities, the relationship pair entity that matches the subject is taken as a problem entity for the target problem. Judging whether a relation pair entity matched with a subject of a target problem exists in relation pair entities contained in an entity relation pair set, wherein the matching refers to the same or similar relation pair entities, and for convenience of matching, the relation pair entities in the entity relation pair can be subjected to paraphrasing or synonym expansion, so that the method is beneficial to quickly determining the problem entity of the target problem, and can be expanded in compatibility, namely, when a user uses a conventional name or an alias of the subject to ask questions, the method can accurately feed back answers to the problems, and the answers cannot be returned because the alias is not recorded in the entity relation pair set.
In addition, if all the relation pairs contained in the subject and entity relation pair set of the target question are not matched, the fact that the answer of the target question is not recorded in the entity relation pair set is explained, corpus expansion and entity relation pair mining are needed, and entity relation pair data are enriched, so that the coverage rate and accuracy rate of answer recall are ensured.
S305: performing problem cluster pattern matching on the target problem based on a problem cluster pattern set, and determining a target problem cluster pattern corresponding to the target problem, wherein the problem cluster pattern set comprises at least one problem cluster pattern, and the problem cluster pattern is obtained by performing frequent sequence pattern mining on a problem set sample.
Specifically, each question cluster pattern in the question cluster pattern set is compared with the target question, and the question cluster pattern included in the target question is determined to be the target question cluster pattern corresponding to the target question.
S307: and determining a target problem cluster type corresponding to the target problem cluster mode according to the mapping relation between the problem cluster mode and the problem cluster type.
Specifically, a problem cluster type corresponding to a problem cluster pattern matched with a target problem cluster pattern is obtained, and the obtained problem cluster type is used as the target problem cluster pattern corresponding to the target problem.
S309: when a preset relationship is formed between the target problem cluster type and the entity relationship in the target entity relationship pair, generating an answer to the target problem according to the relationship in the target entity relationship pair.
In a possible embodiment, when a specific reference relationship or an upper-lower relationship is formed between the target problem cluster type and the entity relationship in the target entity relationship pair, two relationship pair entities in the target entity relationship pair are obtained; removing the relationship pair entity which is the same as the problem entity from the two relationship pair entities to obtain a target relationship pair entity; and generating an answer to the target question for the entity according to the target relation.
Specifically, based on the entity relation pair set and the mapping relation between the problem cluster mode and the problem cluster type, for the target entity relation pair matched with the target problem, judging whether the entity relation in the target problem cluster type and the target entity relation pair forms a specific reference or an upper-lower relation, and if so, answering the problem. For example, for the 6 example questions in Table (2), the results of the various steps mined by the method of the invention are shown in Table (12):
watch (12)
Compared with the prior art, the question-answering processing method provided by the embodiment of the invention has the advantages that:
1. the method forms the question cluster mode based on the frequent sequence of the question mining, so that the question cluster mode is marked with answer type words of the locating question clusters, each question does not need to be identified, the answer type word identification efficiency of the question is greatly improved, meanwhile, the question clusters containing the same sequence mode have a common mode, a mapping relation can be formed with the answer clusters better, and the answer accuracy is improved;
2. according to the method, entity relation pairs are mined based on corpus such as hundred-degree vocabulary entries, if the problem simultaneously satisfies that the problem entity is matched with the relation pairs, and the relation of the entity relation pairs and the type of the problem cluster form a specific reference or upper-lower relation, the entity of the answer can be accurately matched, the question and answer is carried out by combining the problem mode and the entity relation, and the aim of answer recall is more definite;
3. the invention provides a question-answering system construction method for fusing a question cluster mode and an entity relation pair for the first time, the question-answering is carried out by fusing the question cluster mode and the entity relation pair, the efficiency of recall from accuracy and model is relatively high, the method has good operability in industry, the method can be widely applied to the fields of question-answering system construction, knowledge-based answering and the like, for example, in intelligent customer service, knowledge-base question-answering related application scenes, the question cluster mapped based on the question search relation pair and the contained sequence mode is used for accurately locking entity words corresponding to questions as answer recall, and the recall accuracy and answer efficiency are improved.
The embodiment of the invention also provides a question-answering processing device, which can be disposed at a server, and fig. 6 is a schematic structural diagram of the question-answering processing device provided by the embodiment of the invention, please refer to fig. 6, where the device includes a question obtaining module 610, an entity relationship pair matching module 620, a question cluster pattern matching module 630, a question cluster type determining module 640 and an answer determining module 650.
The problem obtaining module 610 is configured to obtain a target problem.
The entity relationship pair matching module 620 is configured to perform entity relationship pair matching on the target problem according to an entity relationship pair set, and determine a target entity relationship pair corresponding to the target problem, where the entity relationship pair set includes at least one entity relationship pair, and each entity relationship pair includes two relationship pair entities and an entity relationship between the two relationship pair entities.
The problem cluster pattern matching module 630 is configured to perform problem cluster pattern matching on the target problem based on a problem cluster pattern set, determine a target problem cluster pattern corresponding to the target problem, where the problem cluster pattern set includes at least one problem cluster pattern, and the problem cluster pattern is obtained by performing frequent sequence pattern mining on a problem set sample.
And a problem cluster type determining module 640, configured to determine a target problem cluster type corresponding to the target problem cluster mode according to a mapping relationship between the problem cluster mode and the problem cluster type.
And the answer determining module 650 is configured to generate an answer to the target question according to the relationship pair entity in the target entity relationship pair when a preset relationship is formed between the target question cluster type and the entity relationship in the target entity relationship pair.
The question-answering processing apparatus in this embodiment is based on the same inventive concept as the method embodiments corresponding to fig. 3-5.
The embodiment of the invention provides a question-answering processing device integrating a question cluster mode and an entity relation pair, which marks the answer type of the question cluster mode based on a question mining question cluster sequence mode, so that the answer type of the question can be accurately identified, and meanwhile, the established relation pair is utilized to carry out entity relation association and entity matching, so that the search result of the question can be more accurately positioned, and the answer of the question can be given. The scheme has wide application value and reference significance in related applications of question and answer system construction, can be widely applied to the fields of question and answer system construction, knowledge preemption and the like, for example, in a knowledge base question and answer related application scene, a question cluster mapped based on a question searching relation pair and a sequence mode is included, so that entity words corresponding to the questions are accurately locked and used as answer recalls; in application scenes such as intelligent customer service, the accuracy and the answering efficiency of recall are improved by combining the entity relation and the question cluster mode constructed by the domain knowledge base. In addition, the application related to the question and answer system belongs to the potential application scene of the invention.
The embodiment of the invention provides electronic equipment, which comprises a processor and a memory, wherein at least one instruction and at least one section of program are stored in the memory, and the at least one instruction or the at least one section of program is loaded and executed by the processor to realize a question-answering processing method corresponding to the method shown in figures 3-5.
The memory may be used to store software programs and modules that the processor executes to perform various functional applications and data processing by executing the software programs and modules stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, application programs required for functions, and the like; the storage data area may store data created according to the use of the device, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory may also include a memory controller to provide access to the memory by the processor.
The embodiment of the invention also provides a schematic structure of a server, referring to fig. 7, the server 700 is used for implementing the question-answering processing method provided in the above embodiment, and specifically, the server structure may include the question-answering processing device. The server 700 may vary considerably in configuration or performance and may include one or more central processing units (Central Processing Units, CPU) 710 (e.g., one or more processors) and memory 730, one or more storage media 720 (e.g., one or more mass storage devices) storing applications 723 or data 722. Wherein memory 730 and storage medium 720 may be transitory or persistent. The program stored in the storage medium 720 may include one or more modules, each of which may include a series of instruction operations on the server. Still further, the central processor 710 may be configured to communicate with the storage medium 720 and execute a series of instruction operations in the storage medium 720 on the server 700. The server 700 may also include one or more power supplies 760, one or more wired or wireless network interfaces 750, one or more input/output interfaces 740, and/or one or more operating systems 721, such as Windows ServerTM, mac OS XTM, unixTM, linuxTM, freeBSDTM, and the like.
Embodiments of the present invention also provide a computer readable storage medium that may be provided in a server to store at least one instruction and at least one program related to a question-answer processing method for implementing a method embodiment, where the at least one instruction and the at least one program are loaded and executed by the processor to implement a question-answer processing method corresponding to fig. 3-5.
Alternatively, in this embodiment, the storage medium may be located in at least one network server among a plurality of network servers of the computer network. Alternatively, in the present embodiment, the storage medium may include, but is not limited to: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. And the foregoing description has been directed to specific embodiments of this specification. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for the device and server embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and references to the parts of the description of the method embodiments are only required.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program for instructing relevant hardware, where the program may be stored in a computer readable storage medium, and the storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The foregoing description of the preferred embodiments of the invention is not intended to limit the invention to the precise form disclosed, and any such modifications, equivalents, and alternatives falling within the spirit and scope of the invention are intended to be included within the scope of the invention.
Claims (11)
1. A question-answering processing method, characterized by comprising:
acquiring a target problem;
performing entity relation pair matching on the target problem according to an entity relation pair set, and determining a target entity relation pair corresponding to the target problem, wherein the entity relation pair set comprises at least one entity relation pair, and each entity relation pair comprises two relation pair entities and an entity relation between the two relation pair entities; one relation pair entity exists in the target entity relation pair, and the relation pair entity is the same as or similar to the words contained in the target problem;
Performing problem cluster pattern matching on the target problem based on a problem cluster pattern set, and determining a target problem cluster pattern corresponding to the target problem, wherein the problem cluster pattern set comprises at least one problem cluster pattern;
determining a target problem cluster type corresponding to the target problem cluster mode according to the mapping relation between the problem cluster mode and the problem cluster type;
when a preset relationship is formed between the target problem cluster type and the entity relationship in the target entity relationship pair, generating an answer to the target problem according to the relationship in the target entity relationship pair; the preset relationship comprises a specific reference relationship or an upper-lower relationship.
2. The method according to claim 1, wherein the performing entity relationship pair matching on the target problem according to the entity relationship pair set, and determining the target entity relationship pair corresponding to the target problem includes:
performing entity matching on the target problem by the entity according to the relation contained in the entity relation pair set to obtain a problem entity corresponding to the target problem;
and determining a target entity relation pair corresponding to the problem entity according to the subordinate relation between each relation pair entity and the entity relation pair in the entity relation pair set.
3. The method of claim 1, wherein the performing the problem cluster pattern matching on the target problem based on the problem cluster pattern set, and determining the target problem cluster pattern corresponding to the target problem, comprises:
comparing each question cluster mode in the question cluster mode set with the target question, and determining the question cluster mode contained in the target question as a target question cluster mode corresponding to the target question.
4. The method according to claim 1, wherein when a preset relationship is formed between the target problem cluster type and the entity relationship in the target entity relationship pair, generating an answer to the target problem according to the relationship in the target entity relationship pair includes:
when a specific reference relationship or an upper-lower relationship is formed between the target problem cluster type and the entity relationship in the target entity relationship pair, acquiring two relationship pair entities in the target entity relationship pair;
removing the relationship pair entity which is the same as the problem entity from the two relationship pair entities to obtain a target relationship pair entity;
and generating an answer to the target question for the entity according to the target relation.
5. The method of claim 1, wherein the generating of the set of problem cluster patterns comprises:
acquiring a question set sample, wherein the question set sample comprises at least one question;
taking a word sequence as a frequent sequence pattern mining object, mining frequent word sequence patterns of each length meeting the minimum support requirement in the problem set sample, and obtaining a first problem pattern set, wherein the first problem pattern set comprises at least two problem sequence patterns;
judging the inclusion relation of the problem sequence modes in the first problem mode set, and determining the category of each problem sequence mode, wherein the category comprises a subsequence and a supersequence;
removing the problem sequence modes with the sub-sequence categories in the first problem mode set to obtain a second problem mode set consisting of the problem sequence modes with the super-sequence categories;
filtering each problem sequence mode in the second problem mode set, and taking the problem sequence mode obtained by the filtering process as a problem cluster mode;
and constructing a problem cluster mode set according to the filtered problem cluster modes.
6. The method of claim 1, wherein prior to the obtaining the target problem, further comprising:
Acquiring a corpus, wherein the corpus comprises at least one corpus;
the following processing is performed for each corpus:
acquiring entities contained in the corpus to obtain an entity set, wherein the entity set comprises at least two entities;
combining the entities in the entity set in pairs to form at least one entity pair, wherein each entity pair comprises two entities;
mining the relation between two entities in each entity pair according to the corpus to obtain an entity relation corresponding to each entity pair, and forming an entity relation pair according to the entity pair and the entity relation corresponding to the entity pair;
and constructing an entity relation pair set based on the composed entity relation pairs.
7. The method of claim 6, wherein mining the relationship between two entities in each entity pair according to the corpus to obtain the entity relationship corresponding to each entity pair, and forming the entity relationship pair according to the entity pair and the entity relationship corresponding to the entity pair, comprises:
counting word frequencies of words in the context where two entities in the entity relation pair are located together;
when word frequency of a word exceeds a set threshold, taking the two entities as relationship pair entities, and taking the word as an entity relationship between the two relationship pair entities;
And establishing a mapping relation between the entity relation and the two relation pair entities to form an entity relation pair.
8. The method of claim 1 or 5, further comprising, prior to the acquiring the target question:
determining a problem cluster type of each problem cluster mode in the problem cluster mode set;
and establishing a mapping relation between the problem cluster type and the problem cluster mode.
9. A question-answering processing apparatus, comprising:
the problem acquisition module is used for acquiring target problems;
the entity relation pair matching module is used for carrying out entity relation pair matching on the target problem according to an entity relation pair set, determining a target entity relation pair corresponding to the target problem, wherein the entity relation pair set comprises at least one entity relation pair, and each entity relation pair comprises two relation pair entities and entity relations between the two relation pair entities; one relation pair entity exists in the target entity relation pair, and the relation pair entity is the same as or similar to the words contained in the target problem;
the problem cluster pattern matching module is used for carrying out problem cluster pattern matching on the target problem based on a problem cluster pattern set, and determining a target problem cluster pattern corresponding to the target problem, wherein the problem cluster pattern set comprises at least one problem cluster pattern;
The problem cluster type determining module is used for determining a target problem cluster type corresponding to the target problem cluster mode according to the mapping relation between the problem cluster mode and the problem cluster type;
the answer determining module is used for generating an answer to the target question for an entity according to the relation in the target entity relation pair when a preset relation is formed between the target question cluster type and the entity relation in the target entity relation pair; the preset relationship comprises a specific reference relationship or an upper-lower relationship.
10. A computer readable storage medium having stored therein at least one instruction or at least one program loaded and executed by a processor to implement a question-answering method according to any one of claims 1 to 7.
11. An electronic device comprising a processor and a memory, wherein the memory stores at least one instruction and at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the question-answer processing method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010529247.8A CN111625640B (en) | 2020-06-11 | 2020-06-11 | Question and answer processing method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010529247.8A CN111625640B (en) | 2020-06-11 | 2020-06-11 | Question and answer processing method, device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111625640A CN111625640A (en) | 2020-09-04 |
| CN111625640B true CN111625640B (en) | 2023-11-14 |
Family
ID=72272244
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010529247.8A Active CN111625640B (en) | 2020-06-11 | 2020-06-11 | Question and answer processing method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111625640B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112131353A (en) * | 2020-10-12 | 2020-12-25 | 吉林大学 | Method for solving full text interpretation analysis problem |
| CN116595156B (en) * | 2023-07-17 | 2023-12-19 | 北京普华恒信技术服务有限公司 | Knowledge management system and knowledge management method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103810218A (en) * | 2012-11-14 | 2014-05-21 | 北京百度网讯科技有限公司 | Problem cluster-based automatic asking and answering method and device |
| CN105005564A (en) * | 2014-04-17 | 2015-10-28 | 北京搜狗科技发展有限公司 | Data processing method and apparatus based on question-and-answer platform |
| CN106997399A (en) * | 2017-05-24 | 2017-08-01 | 海南大学 | A kind of classification question answering system design method that framework is associated based on data collection of illustrative plates, Information Atlas, knowledge mapping and wisdom collection of illustrative plates |
| CN108536852A (en) * | 2018-04-16 | 2018-09-14 | 上海智臻智能网络科技股份有限公司 | Question and answer exchange method and device, computer equipment and computer readable storage medium |
| CN109284363A (en) * | 2018-12-03 | 2019-01-29 | 北京羽扇智信息科技有限公司 | A kind of answering method, device, electronic equipment and storage medium |
| CN110019149A (en) * | 2019-01-30 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of method for building up of service knowledge base, device and equipment |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104598445B (en) * | 2013-11-01 | 2019-05-10 | 腾讯科技（深圳）有限公司 | Automatically request-answering system and method |
-
2020
- 2020-06-11 CN CN202010529247.8A patent/CN111625640B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103810218A (en) * | 2012-11-14 | 2014-05-21 | 北京百度网讯科技有限公司 | Problem cluster-based automatic asking and answering method and device |
| CN105005564A (en) * | 2014-04-17 | 2015-10-28 | 北京搜狗科技发展有限公司 | Data processing method and apparatus based on question-and-answer platform |
| CN106997399A (en) * | 2017-05-24 | 2017-08-01 | 海南大学 | A kind of classification question answering system design method that framework is associated based on data collection of illustrative plates, Information Atlas, knowledge mapping and wisdom collection of illustrative plates |
| CN108536852A (en) * | 2018-04-16 | 2018-09-14 | 上海智臻智能网络科技股份有限公司 | Question and answer exchange method and device, computer equipment and computer readable storage medium |
| CN109284363A (en) * | 2018-12-03 | 2019-01-29 | 北京羽扇智信息科技有限公司 | A kind of answering method, device, electronic equipment and storage medium |
| CN110019149A (en) * | 2019-01-30 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of method for building up of service knowledge base, device and equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111625640A (en) | 2020-09-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109299110B (en) | Data query method and device, storage medium and electronic equipment | |
| CN110704743B (en) | Semantic search method and device based on knowledge graph | |
| CN105359144B (en) | Natural Language Search result for intent query | |
| US7962486B2 (en) | Method and system for discovery and modification of data cluster and synonyms | |
| CN107590214B (en) | Recommendation method and device for search keywords and electronic equipment | |
| CN111625640B (en) | Question and answer processing method, device and storage medium | |
| US10268655B2 (en) | Method, device, server and storage medium of searching a group based on social network | |
| JP2005108221A (en) | System and method for indexing query, rule and subscription | |
| CN110674394B (en) | Knowledge graph-based information recommendation method, device and storage medium | |
| CN110209809B (en) | Text clustering method and device, storage medium and electronic device | |
| US9424344B2 (en) | Method and apparatus for natural language search for variables | |
| CN112507176A (en) | Automatic determination method and device for domain name infringement, electronic equipment and storage medium | |
| CN109885651A (en) | A kind of question pushing method and device | |
| JP2007249578A (en) | Attribute presumption program and attribute information providing system | |
| CN110245357B (en) | Main entity identification method and device | |
| JP5224453B2 (en) | Geographic feature information extraction method and system | |
| CN105512270B (en) | Method and device for determining related objects | |
| CN108509449B (en) | Information processing method and server | |
| CN114610859A (en) | Product recommendation method, device and equipment based on content and collaborative filtering | |
| JP7168334B2 (en) | Information processing device, information processing method and program | |
| WO2020071666A1 (en) | Method for generating chatbot statements by using frequency table | |
| CN110633430A (en) | Event discovery method, device, equipment and computer readable storage medium | |
| CN109766415B (en) | Book directory positioning method and system | |
| CN116738065B (en) | Enterprise searching method, device, equipment and storage medium | |
| CN110659345B (en) | Data pushing method, device and equipment for fact report and storage medium |
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 |