CN107305550A - A kind of intelligent answer method and device - Google Patents
A kind of intelligent answer method and device Download PDFInfo
- Publication number
- CN107305550A CN107305550A CN201610244036.3A CN201610244036A CN107305550A CN 107305550 A CN107305550 A CN 107305550A CN 201610244036 A CN201610244036 A CN 201610244036A CN 107305550 A CN107305550 A CN 107305550A
- Authority
- CN
- China
- Prior art keywords
- question
- current
- omission
- current question
- complete
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
-
- 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
- G06F16/3332—Query translation
- G06F16/3335—Syntactic pre-processing, e.g. stopword elimination, stemming
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of intelligent answer method and device.This method includes：Current question sentence is obtained, the current question sentence includes the question sentence that user currently inputs；Whether according to the current question sentence, it is incomplete question sentence to determine the current question sentence, and the incomplete question sentence, which exists, omits word, and the omission word includes lacking the keyword for retrieving the current question sentence answer；If it is determined that the current question sentence is incomplete question sentence, then the current question sentence is subjected to integrality recovery operation, to obtain complete question sentence, the complete question sentence includes the question sentence for omitting word is not present；According to the complete question sentence, the answer of the complete question sentence is retrieved in database.The integrality recovery to incomplete question sentence is realized, so as to retrieve accurate answer in database.
Description
Technical Field
The invention relates to computer technology, in particular to an intelligent question answering method and device.
Background
With the continuous development of computer technology, users can intelligently ask and answer with a computer, that is, in a ask-answer form, the computer can accurately position answers of questions asked by the users.
In the existing intelligent question-answering system, a user presents a question to a computer, the computer can correctly select an answer to the question presented by the user from a knowledge base, then the user presents a new question to the computer, and the computer continues to correctly select an answer to the new question presented by the user from the knowledge base.
However, the new question posed by the user may be an incomplete question, e.g., what is the 4008 service? The second problem is: how to do it? The computer will not be able to process the incomplete question, resulting in a failure to provide the user with an accurate answer.
Disclosure of Invention
In order to solve the technical problems, the invention provides an intelligent question answering method and device, which are used for solving the problem that a computer cannot provide accurate answers for users.
In order to achieve the purpose of the invention, the invention provides an intelligent question answering method, which comprises the following steps:
acquiring a current question, wherein the current question comprises a question currently input by a user;
determining whether the current question is an incomplete question or not according to the current question, wherein the incomplete question has omission words which comprise key words lacking answers for retrieving the current question;
if the current question is determined to be an incomplete question, performing integrity recovery operation on the current question to obtain an integral question, wherein the integral question comprises a question without omitted words;
and searching answers of the complete question in a database according to the complete question.
Further, before the performing an integrity recovery operation on the current question to obtain a complete question, the method further includes:
acquiring at least one omission judgment result feature vector of the omitted words in the current question, wherein the omission judgment result feature vector comprises a feature vector obtained according to any one of the following information or a combination thereof: an omission type, a pronoun type, an omission component position, an omission component grammar role.
Further, before the performing an integrity recovery operation on the current question to obtain a complete question, the method further includes:
acquiring at least one first keyword of a previous question, wherein the previous question comprises a question acquired before the current question and input by the same user identifier as the current question;
according to the first keyword, obtaining a omitting candidate word characteristic value vector, wherein the omitting candidate word characteristic value vector comprises a characteristic vector obtained according to any one or a combination of the following information: the entity type, the grammatical role of the entity in the previous question, and the interval distance, wherein the interval distance comprises the distance between the previous question where the first keyword is located and the current question.
Further, the performing integrity recovery operation on the current question to obtain a complete question further includes:
multiplying the omission candidate word feature value vector by the omission judgment result feature vector to obtain a multiplication result;
according to the multiplication result, determining a score of the multiplication result corresponding to a first list;
sorting all scores in a descending order, and determining keywords corresponding to the first N scores as candidate keywords;
supplementing the candidate keywords to the current question to obtain a recovery question;
carrying out syntactic analysis on the recovery question sentence to obtain a scoring result of the syntactic analysis;
and according to the scoring result, taking the recovery question with the highest scoring result as the complete question of the current question.
Further, before the obtaining of the at least one first keyword of the previous question sentence, the method further includes:
determining whether the previous question and the current question have relevance or not;
if the correlation exists, executing at least one first keyword of the question sentence before obtaining;
and if the correlation does not exist, searching the answer of the current question in the database.
The invention also provides an intelligent question answering device, which comprises:
the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring a current question, and the current question comprises a question currently input by a user;
a determining module, configured to determine whether the current question is an incomplete question according to the current question, where the incomplete question includes an omitted word, and the omitted word includes a keyword lacking an answer for retrieving the current question;
the integrity recovery module is used for carrying out integrity recovery operation on the current question to obtain a complete question if the current question is determined to be an incomplete question, wherein the complete question comprises a question without omitted words;
and the retrieval module is used for retrieving answers of the complete question in a database according to the complete question.
Further, the obtaining module is specifically configured to obtain at least one omission judgment result feature vector of the omitted word in the current question, where the omission judgment result feature vector includes a feature vector obtained according to any one of the following information or a combination thereof: an omission type, a pronoun type, an omission component position, an omission component grammar role.
Further, the obtaining module is further configured to obtain at least one first keyword of a previous question, where the previous question includes a question obtained before the current question and input by the same user identifier as the current question; according to the first keyword, obtaining a omitting candidate word characteristic value vector, wherein the omitting candidate word characteristic value vector comprises a characteristic vector obtained according to any one or a combination of the following information: the entity type, the grammatical role of the entity in the previous question, and the interval distance, wherein the interval distance comprises the distance between the previous question where the first keyword is located and the current question.
Further, the integrity recovery module is further configured to multiply the omitted candidate word feature value vector with the omitted determination result feature vector to obtain a multiplication result; according to the multiplication result, determining a score of the multiplication result corresponding to a first list; sorting all scores in a descending order, and determining keywords corresponding to the first N scores as candidate keywords; supplementing the candidate keywords to the current question to obtain a recovery question; carrying out syntactic analysis on the recovery question sentence to obtain a scoring result of the syntactic analysis; and according to the scoring result, taking the recovery question with the highest scoring result as the complete question of the current question.
Further, the method also comprises the following steps: a processing module;
the processing module is used for determining whether the correlation exists between the previous question and the current question; if the correlation exists, executing at least one first keyword of the question sentence before obtaining; and if the correlation does not exist, searching the answer of the current question in the database.
Compared with the prior art, the invention comprises the following steps: acquiring a current question, wherein the current question comprises a question currently input by a user; determining whether the current question is an incomplete question or not according to the current question, wherein the incomplete question has omission words which comprise key words lacking answers for retrieving the current question; if the current question is determined to be an incomplete question, performing integrity recovery operation on the current question to obtain an integral question, wherein the integral question comprises a question without omitted words; and searching answers of the complete question in a database according to the complete question. The integrity recovery of the incomplete question is realized, so that an accurate answer can be retrieved in a database.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the example serve to explain the principles of the invention and not to limit the invention.
FIG. 1 is a schematic flow chart diagram of an embodiment of an intelligent question answering method according to the present invention;
FIG. 2 is a schematic structural diagram of an embodiment of an intelligent question answering device according to the present invention;
fig. 3 is a schematic structural diagram of a second embodiment of the intelligent question answering device according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that the embodiments and features of the embodiments in the present application may be arbitrarily combined with each other without conflict.
The intelligent question answering method provided by the embodiment of the invention can be particularly applied to the intelligent answer provided by a computer. The intelligent question-answering method provided by the embodiment can be specifically executed through an intelligent question-answering device, the intelligent question-answering device can be integrated in a mobile terminal, a computer or be separately arranged, and the intelligent question-answering device can be realized in a software and/or hardware mode. The following describes the intelligent question answering method and apparatus provided in this embodiment in detail.
FIG. 1 is a schematic flow chart diagram of an embodiment of an intelligent question answering method according to the present invention; as shown in fig. 1, the executing subject in this embodiment may be an intelligent question answering device, and the intelligent question answering method provided by the present invention includes:
step 101, obtaining a current question.
In this embodiment, the current question includes a question currently input by the user.
And step 102, determining whether the current question is an incomplete question or not according to the current question.
In this embodiment, the incomplete question has an omission word, and the omission word includes a keyword lacking an answer for retrieving the current question.
Specifically, whether the current question has the omission phenomenon is judged, and if the current question has the omission phenomenon, the omission component is recovered according to the content of the current question and the entity recognition result. Omission herein means omission of key components or the use of pronouns for key components. In the omission phenomenon judgment work, linguistic data are labeled firstly, effective features are extracted, an omission judgment model is trained, whether the omission phenomenon exists in words in a sentence is judged by using the omission judgment model, and information such as the position of an omitted component and a grammatical role is judged under the condition that the omission phenomenon exists.
For example, the current question may be determined whether the current question is an incomplete question by omitting the decision model. Wherein, the omitting judgment model training mode comprises the following steps: the multiple question sentences are preprocessed, that is, the parts of speech, the sequence of speech, the omission characteristics, and the like of the words in the question sentences are labeled, for example, the omission characteristics may be the omission subject, the verb, and the like. Then, reading the preprocessed corpus, extracting characteristic values, and performing offline training of the model; the characteristic value comprises pronoun components, omission characteristics, semantic characteristics, entity word information and the like. Then, off-line training is performed to generate an omitted recovery model.
Step 103, if the current question is determined to be an incomplete question, performing integrity recovery operation on the current question to obtain an integral question.
In the present embodiment, the complete question includes a question in which the omitted word does not exist.
Specifically, after the omission component position is determined, it is the most important to determine the restoration content next.
And recovering the candidate words into the first few question sentences of the same user, and extracting one or more entity words through entity recognition. And selecting the best candidate word as the omitted recovery word through algorithm calculation and syntactic analysis.
And step 104, searching answers of the complete question in a database according to the complete question.
It should be noted that the database may be a conventional relational database, or alternatively, an indexed database, and is not limited herein.
In the embodiment, a current question is obtained, wherein the current question comprises a question currently input by a user; determining whether the current question is an incomplete question or not according to the current question, wherein the incomplete question has omission words which comprise key words lacking answers for retrieving the current question; if the current question is determined to be an incomplete question, performing integrity recovery operation on the current question to obtain an integral question, wherein the integral question comprises a question without omitted words; and searching answers of the complete question in a database according to the complete question. The integrity recovery of the incomplete question is realized, so that an accurate answer can be retrieved in a database.
On the basis of the above embodiment, the determining, according to the current question, whether the current question is an incomplete question further includes:
and if the current question is determined not to be an incomplete question, searching the answer of the current question in a database.
On the basis of the foregoing embodiment, before performing an integrity recovery operation on the current question to obtain a complete question, the method further includes:
acquiring at least one omission judgment result feature vector of the omitted words in the current question, wherein the omission judgment result feature vector comprises a feature vector obtained according to any one of the following information or a combination thereof: an omission type, a pronoun type, an omission component position, an omission component grammar role.
On the basis of the foregoing embodiment, before performing an integrity recovery operation on the current question to obtain a complete question, the method further includes:
acquiring at least one first keyword of a previous question, wherein the previous question comprises a question acquired before the current question and input by the same user identifier as the current question;
according to the first keyword, obtaining a omitting candidate word characteristic value vector, wherein the omitting candidate word characteristic value vector comprises a characteristic vector obtained according to any one or a combination of the following information: the entity type, the grammatical role of the entity in the previous question, and the interval distance, wherein the interval distance comprises the distance between the previous question where the first keyword is located and the current question.
Further, on the basis of the foregoing embodiment, the performing integrity recovery operation on the current question to obtain a complete question further includes:
multiplying the omission candidate word feature value vector by the omission judgment result feature vector to obtain a multiplication result;
according to the multiplication result, determining a score of the multiplication result corresponding to a first list;
sorting all scores in a descending order, and determining keywords corresponding to the first N scores as candidate keywords;
supplementing the candidate keywords to the current question to obtain a recovery question;
carrying out syntactic analysis on the recovery question sentence to obtain a scoring result of the syntactic analysis;
and according to the scoring result, taking the recovery question with the highest scoring result as the complete question of the current question.
It should be noted that, before acquiring at least one first keyword of a preceding question, the method further includes:
determining whether the previous question and the current question have relevance or not;
if the correlation exists, executing at least one first keyword of the question sentence before obtaining;
and if the correlation does not exist, searching the answer of the current question in the database.
For example, the User: what are the 4G set of meals enjoyed? Intelligent customer service IQA: the enjoy 4G package is a certain package in China telecom. Next, the User: how to do it? Before answering, the intelligent customer service IQA can do the following operations:
step 1, preparing and marking domain linguistic data, and training a model in an off-line mode. Respectively training a context recognition model, an entity recognition model and an omission judgment model;
step 2, judging whether the user has cache, if so, continuing to execute context recognition in the step 3, otherwise, executing search in the step 7; the present example has a cache;
step 3, calling a context recognition model to judge whether the two contexts are the same, if so, continuing to execute step 4 to omit judgment, otherwise, executing step 7 to search; this example is the same context;
and 4, calling an omission judgment model to judge whether omission exists or not, if so, determining the omission position and the omission components, obtaining the omission judgment result feature vector A, and continuing to execute the entity identification in the step 5. Otherwise, executing step 7 search; in this example, there is an omission, and there is a missing subject, the omission is located before "what" and
step 5, calling an entity recognition model to recognize an entity which can be used for omitting recovery; continuing to execute step 6 to omit recovery; the entity recognition model of the embodiment recognizes the previous question entity 'Lexiang 4G package'
Step 6, extracting a group of vectors of feature values of omitted candidate words for each entity, wherein the vectors comprise entity categories, grammatical roles of the entities in original sentences, interval distances and the like, the interval distances refer to distances between the entities and current question sentences, and the question sentences are used as distance units; and calculating the score of the candidate entity word by omitting the candidate word feature value vector, and selecting the recovery word according to the score. The omission judgment result feature vector A obtained in the step 4 is a weight coefficient of the candidate word feature vector; the entity identified in this example is the subject in the previous question, and is spaced from the current question by a distance of 1, and is calculated as the best recovery word; after recovery, the current question is: how do a 4G set of fun?
Step 7, searching the corpus, performing result post-processing, screening query results (if a plurality of query results are hit, selecting the best answer by using similarity calculation, wherein the part is not in the scope of the invention) and caching the query conditions of the user;
and 8, returning the final result to the user.
For another example, the embodiment is applicable to a life service APP. User: how do the weather in Beijing? Intelligent customer service IQA: weather conditions in Beijing: cloudy turns fine. User: nanjing woolen? Before answering, the intelligent customer service IQA can do the following operations:
step 1, preparing and marking domain linguistic data, and training a model in an off-line mode. Respectively training a context recognition model, an entity recognition model and an omission judgment model;
step 2, judging whether the user has cache, if so, continuing to execute context recognition in the step 3, otherwise, executing search in the step 7; the present example has a cache;
step 3, calling a context recognition model to judge whether the two contexts are the same, if so, continuing to execute step 4 to omit judgment, otherwise, executing step 7 to search; this example is the same context;
and 4, calling an omission judgment model to judge whether omission exists or not, if so, determining the omission position and the omission components, obtaining the omission judgment result feature vector A, and continuing to execute the entity identification in the step 5. Otherwise, executing step 7 search; the example has omission, and has regional word subject "Nanjing", no object, and omission position behind the subject
Step 5, calling an entity recognition model to recognize an entity which can be used for omitting recovery; continuing to execute step 6 to omit recovery; the entity recognition model of the embodiment recognizes that the entities of the previous question have Nanjing and weather "
Step 6, extracting a group of vectors of feature values of omitted candidate words for each entity, wherein the vectors comprise entity types, grammatical roles of the entities in original sentences, interval distances (the distance between the entities and the current question is taken as a distance unit), and the like; and calculating the score of the candidate entity word by omitting the candidate word feature value vector, and selecting the recovery word according to the score. The omission judgment result feature vector A obtained in the step 4 is a weight coefficient of the candidate word feature vector; the entity identified in the example is the subject and the object in the previous question, the interval distance between the entity and the current question is 1, the current question lacks the object, and the weather is calculated as the best recovery word; after recovery, the current question is: nanjing weather woollen?
Step 7, searching the corpus, performing result post-processing, screening query results (if a plurality of query results are hit, selecting the best answer by using similarity calculation, wherein the part is not in the scope of the invention) and caching the query conditions of the user;
and 8, returning the final result to the user.
FIG. 2 is a schematic structural diagram of an embodiment of an intelligent question answering device according to the present invention; as shown in fig. 2, the intelligent question answering device provided by the present invention includes: an acquisition module 21, a determination module 22, an integrity restoration module 23 and a retrieval module 24. Wherein,
an obtaining module 21, configured to obtain a current question, where the current question includes a question currently input by a user;
a determining module 22, configured to determine, according to the current question, whether the current question is an incomplete question, where an omitted word exists in the incomplete question, and the omitted word includes a keyword lacking an answer for retrieving the current question;
the integrity recovery module 23 is configured to, if it is determined that the current question is an incomplete question, perform integrity recovery operation on the current question to obtain an complete question, where the complete question includes a question without an omitted word;
and the retrieval module 24 is used for retrieving answers of the complete question from a database according to the complete question.
In the embodiment, a current question is obtained, wherein the current question comprises a question currently input by a user; determining whether the current question is an incomplete question or not according to the current question, wherein the incomplete question has omission words which comprise key words lacking answers for retrieving the current question; if the current question is determined to be an incomplete question, performing integrity recovery operation on the current question to obtain an integral question, wherein the integral question comprises a question without omitted words; and searching answers of the complete question in a database according to the complete question. The integrity recovery of the incomplete question is realized, so that an accurate answer can be retrieved in a database.
Further, on the basis of the foregoing embodiment, the obtaining module 21 is specifically configured to obtain at least one omission judgment result feature vector of the omitted word in the current question, where the omission judgment result feature vector includes a feature vector obtained according to any one of the following information or a combination thereof: an omission type, a pronoun type, an omission component position, an omission component grammar role.
Further, on the basis of the foregoing embodiment, the obtaining module 21 is further configured to obtain at least one first keyword of a previous question, where the previous question includes a question that is obtained before the current question and is input by the same user identifier as the current question; according to the first keyword, obtaining a omitting candidate word characteristic value vector, wherein the omitting candidate word characteristic value vector comprises a characteristic vector obtained according to any one or a combination of the following information: the entity type, the grammatical role of the entity in the previous question, and the interval distance, wherein the interval distance comprises the distance between the previous question where the first keyword is located and the current question.
Further, on the basis of the foregoing embodiment, the integrity recovering module 23 is further configured to multiply the omitted candidate word feature value vector by the omitted determination result feature vector to obtain a multiplication result; according to the multiplication result, determining a score of the multiplication result corresponding to a first list; sorting all scores in a descending order, and determining keywords corresponding to the first N scores as candidate keywords; supplementing the candidate keywords to the current question to obtain a recovery question; carrying out syntactic analysis on the recovery question sentence to obtain a scoring result of the syntactic analysis; and according to the scoring result, taking the recovery question with the highest scoring result as the complete question of the current question.
FIG. 3 is a schematic structural diagram of a second embodiment of the intelligent question answering device according to the present invention; as shown in fig. 3, on the basis of the above embodiment, the intelligent question answering device provided by the present invention may further include: a processing module 25;
the processing module 25 is configured to determine whether there is a correlation between the previous question and the current question; if the correlation exists, executing at least one first keyword of the question sentence before obtaining; and if the correlation does not exist, searching the answer of the current question in the database.
In the embodiment, the completeness recovery of the incomplete question is realized, and the entity composed of multiple words is obtained, so that an accurate answer can be retrieved from the database, and the accuracy of the answer is improved. Although the embodiments of the present invention have been described above, the above description is only for the convenience of understanding the present invention, and is not intended to limit the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. An intelligent question answering method is characterized by comprising the following steps:
acquiring a current question, wherein the current question comprises a question currently input by a user;
determining whether the current question is an incomplete question or not according to the current question, wherein the incomplete question has omission words which comprise key words lacking answers for retrieving the current question;
if the current question is determined to be an incomplete question, performing integrity recovery operation on the current question to obtain an integral question, wherein the integral question comprises a question without omitted words;
and searching answers of the complete question in a database according to the complete question.
2. The method of claim 1, wherein before performing the integrity recovery operation on the current question to obtain the complete question, further comprising:
acquiring at least one omission judgment result feature vector of the omitted words in the current question, wherein the omission judgment result feature vector comprises a feature vector obtained according to any one of the following information or a combination thereof: an omission type, a pronoun type, an omission component position, an omission component grammar role.
3. The method according to claim 1 or 2, wherein before performing the integrity recovery operation on the current question to obtain a complete question, the method further comprises:
acquiring at least one first keyword of a previous question, wherein the previous question comprises a question acquired before the current question and input by the same user identifier as the current question;
according to the first keyword, obtaining a omitting candidate word characteristic value vector, wherein the omitting candidate word characteristic value vector comprises a characteristic vector obtained according to any one or a combination of the following information: the entity type, the grammatical role of the entity in the previous question, and the interval distance, wherein the interval distance comprises the distance between the previous question where the first keyword is located and the current question.
4. The method according to any one of claims 1 to 3, wherein the performing an integrity recovery operation on the current question to obtain a complete question further comprises:
multiplying the omission candidate word feature value vector by the omission judgment result feature vector to obtain a multiplication result;
according to the multiplication result, determining a score of the multiplication result corresponding to a first list;
sorting all scores in a descending order, and determining keywords corresponding to the first N scores as candidate keywords;
supplementing the candidate keywords to the current question to obtain a recovery question;
carrying out syntactic analysis on the recovery question sentence to obtain a scoring result of the syntactic analysis;
and according to the scoring result, taking the recovery question with the highest scoring result as the complete question of the current question.
5. The method according to claims 1-4, wherein said obtaining at least one first keyword of a previous question further comprises:
determining whether the previous question and the current question have relevance or not;
if the correlation exists, executing at least one first keyword of the question sentence before obtaining;
and if the correlation does not exist, searching the answer of the current question in the database.
6. An intelligent question answering device, comprising:
the system comprises an acquisition module, a display module and a display module, wherein the acquisition module is used for acquiring a current question, and the current question comprises a question currently input by a user;
a determining module, configured to determine whether the current question is an incomplete question according to the current question, where the incomplete question includes an omitted word, and the omitted word includes a keyword lacking an answer for retrieving the current question;
the integrity recovery module is used for carrying out integrity recovery operation on the current question to obtain a complete question if the current question is determined to be an incomplete question, wherein the complete question comprises a question without omitted words;
and the retrieval module is used for retrieving answers of the complete question in a database according to the complete question.
7. The apparatus according to claim 6, wherein the obtaining module is specifically configured to obtain at least one omission decision result feature vector of the omitted words in the current question sentence, where the omission decision result feature vector includes a feature vector obtained according to any one or a combination of the following information: an omission type, a pronoun type, an omission component position, an omission component grammar role.
8. The apparatus according to claim 6 or 7, wherein the obtaining module is further configured to obtain at least one first keyword of a previous question, where the previous question includes a question obtained before the current question and input by the same user identifier as the current question; according to the first keyword, obtaining a omitting candidate word characteristic value vector, wherein the omitting candidate word characteristic value vector comprises a characteristic vector obtained according to any one or a combination of the following information: the entity type, the grammatical role of the entity in the previous question, and the interval distance, wherein the interval distance comprises the distance between the previous question where the first keyword is located and the current question.
9. The apparatus according to any one of claims 6 to 8, wherein the integrity recovery module is further configured to multiply the omission candidate word feature value vector by the omission decision result feature vector to obtain a multiplication result; according to the multiplication result, determining a score of the multiplication result corresponding to a first list; sorting all scores in a descending order, and determining keywords corresponding to the first N scores as candidate keywords; supplementing the candidate keywords to the current question to obtain a recovery question; carrying out syntactic analysis on the recovery question sentence to obtain a scoring result of the syntactic analysis; and according to the scoring result, taking the recovery question with the highest scoring result as the complete question of the current question.
10. The apparatus of claims 6-9, further comprising: a processing module;
the processing module is used for determining whether the correlation exists between the previous question and the current question; if the correlation exists, executing at least one first keyword of the question sentence before obtaining; and if the correlation does not exist, searching the answer of the current question in the database.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610244036.3A CN107305550A (en) | 2016-04-19 | 2016-04-19 | A kind of intelligent answer method and device |
| PCT/CN2017/078844 WO2017181834A1 (en) | 2016-04-19 | 2017-03-30 | Intelligent question and answer method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610244036.3A CN107305550A (en) | 2016-04-19 | 2016-04-19 | A kind of intelligent answer method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107305550A true CN107305550A (en) | 2017-10-31 |
Family
ID=60115675
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610244036.3A Pending CN107305550A (en) | 2016-04-19 | 2016-04-19 | A kind of intelligent answer method and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN107305550A (en) |
| WO (1) | WO2017181834A1 (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108170792A (en) * | 2017-12-27 | 2018-06-15 | 北京百度网讯科技有限公司 | Question and answer bootstrap technique, device and computer equipment based on artificial intelligence |
| CN108763356A (en) * | 2018-05-16 | 2018-11-06 | 深圳市三宝创新智能有限公司 | A kind of intelligent robot chat system and method based on the search of similar sentence |
| CN109284503A (en) * | 2018-10-22 | 2019-01-29 | 传神语联网网络科技股份有限公司 | Translate Statement Completion judgment method and system |
| CN109344238A (en) * | 2018-09-18 | 2019-02-15 | 阿里巴巴集团控股有限公司 | The benefit word method and apparatus of user's question sentence |
| CN110334268A (en) * | 2019-07-05 | 2019-10-15 | 北京国创动力文化传媒有限公司 | A kind of block chain project hot word generation method and device |
| CN111507088A (en) * | 2020-04-15 | 2020-08-07 | 深圳前海微众银行股份有限公司 | Statement completion method, device and readable storage medium |
| CN111984766A (en) * | 2019-05-21 | 2020-11-24 | 华为技术有限公司 | Missing semantic completion method and device |
| WO2021012772A1 (en) * | 2019-07-22 | 2021-01-28 | 中兴通讯股份有限公司 | Speech information processing method and device, storage medium, and electronic device |
| CN112328762A (en) * | 2020-11-04 | 2021-02-05 | 平安科技（深圳）有限公司 | Question and answer corpus generation method and device based on text generation model |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109933779A (en) * | 2017-12-18 | 2019-06-25 | 苏宁云商集团股份有限公司 | User's intension recognizing method and system |
| US11423068B2 (en) * | 2018-11-27 | 2022-08-23 | Google Llc | Canonicalizing search queries to natural language questions |
| CN110377803B (en) * | 2019-06-14 | 2023-04-07 | 创新先进技术有限公司 | Information processing method and device |
| CN112988987B (en) * | 2019-12-16 | 2024-06-28 | 科沃斯商用机器人有限公司 | Man-machine interaction method and device, intelligent robot and storage medium |
| CN111563147B (en) * | 2020-04-03 | 2023-09-22 | 北京明略软件系统有限公司 | Entity linking method and device in knowledge question-answering system |
| CN111813916B (en) * | 2020-07-21 | 2024-03-05 | 华润数字科技有限公司 | Intelligent question-answering method, device, computer equipment and medium |
| CN111984763B (en) * | 2020-08-28 | 2023-09-19 | 海信电子科技(武汉)有限公司 | Question answering processing method and intelligent device |
| CN112380836A (en) * | 2020-11-12 | 2021-02-19 | 华侨大学 | Intelligent Chinese message question generating method |
| CN114064874A (en) * | 2021-11-19 | 2022-02-18 | 浙江百应科技有限公司 | Knowledge base problem adding method and device based on vector search engine |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9042540B2 (en) * | 2012-10-30 | 2015-05-26 | Teletech Holdings, Inc. | Method for providing support using answer engine and dialog rules |
| CN103995870A (en) * | 2014-05-21 | 2014-08-20 | 百度在线网络技术（北京）有限公司 | Interactive searching method and device |
| CN105373527B (en) * | 2014-08-27 | 2020-11-27 | 中兴通讯股份有限公司 | An omission recovery method and question answering system |
| CN106294340A (en) * | 2015-05-12 | 2017-01-04 | 中兴通讯股份有限公司 | A kind of intelligent customer service processing method and device thereof |
| CN105159996B (en) * | 2015-09-07 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | Depth question and answer service providing method based on artificial intelligence and device |
| CN105589844B (en) * | 2015-12-18 | 2017-08-08 | 北京中科汇联科技股份有限公司 | It is a kind of to be used to take turns the method for lacking semantic supplement in question answering system more |
-
2016
- 2016-04-19 CN CN201610244036.3A patent/CN107305550A/en active Pending
-
2017
- 2017-03-30 WO PCT/CN2017/078844 patent/WO2017181834A1/en active Application Filing
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108170792B (en) * | 2017-12-27 | 2021-12-28 | 北京百度网讯科技有限公司 | Question and answer guiding method and device based on artificial intelligence and computer equipment |
| CN108170792A (en) * | 2017-12-27 | 2018-06-15 | 北京百度网讯科技有限公司 | Question and answer bootstrap technique, device and computer equipment based on artificial intelligence |
| CN108763356A (en) * | 2018-05-16 | 2018-11-06 | 深圳市三宝创新智能有限公司 | A kind of intelligent robot chat system and method based on the search of similar sentence |
| CN109344238A (en) * | 2018-09-18 | 2019-02-15 | 阿里巴巴集团控股有限公司 | The benefit word method and apparatus of user's question sentence |
| CN109284503A (en) * | 2018-10-22 | 2019-01-29 | 传神语联网网络科技股份有限公司 | Translate Statement Completion judgment method and system |
| CN109284503B (en) * | 2018-10-22 | 2023-08-18 | 传神语联网网络科技股份有限公司 | Translation statement ending judgment method and system |
| US12135941B2 (en) | 2019-05-21 | 2024-11-05 | Huawei Technologies Co., Ltd. | Missing semantics complementing method and apparatus |
| CN111984766A (en) * | 2019-05-21 | 2020-11-24 | 华为技术有限公司 | Missing semantic completion method and device |
| WO2020233380A1 (en) * | 2019-05-21 | 2020-11-26 | 华为技术有限公司 | Missing semantic completion method and apparatus |
| CN111984766B (en) * | 2019-05-21 | 2023-02-24 | 华为技术有限公司 | Missing semantic completion method and device |
| CN110334268A (en) * | 2019-07-05 | 2019-10-15 | 北京国创动力文化传媒有限公司 | A kind of block chain project hot word generation method and device |
| CN110334268B (en) * | 2019-07-05 | 2022-01-14 | 李晨 | Block chain project hot word generation method and device |
| WO2021012772A1 (en) * | 2019-07-22 | 2021-01-28 | 中兴通讯股份有限公司 | Speech information processing method and device, storage medium, and electronic device |
| CN111507088A (en) * | 2020-04-15 | 2020-08-07 | 深圳前海微众银行股份有限公司 | Statement completion method, device and readable storage medium |
| CN112328762A (en) * | 2020-11-04 | 2021-02-05 | 平安科技（深圳）有限公司 | Question and answer corpus generation method and device based on text generation model |
| CN112328762B (en) * | 2020-11-04 | 2023-12-19 | 平安科技（深圳）有限公司 | Question-answer corpus generation method and device based on text generation model |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2017181834A1 (en) | 2017-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107305550A (en) | A kind of intelligent answer method and device | |
| CN110196901B (en) | Method and device for constructing dialog system, computer equipment and storage medium | |
| CN107729468B (en) | Answer extraction method and system based on deep learning | |
| CN107480143B (en) | Method and system for dialogue topic segmentation based on contextual relevance | |
| CN108304372B (en) | Entity extraction method and device, computer equipment and storage medium | |
| US10831769B2 (en) | Search method and device for asking type query based on deep question and answer | |
| EP2833271A1 (en) | Multimedia question and answer system and method | |
| CN105930452A (en) | Smart answering method capable of identifying natural language | |
| CN104050256A (en) | Initiative study-based questioning and answering method and questioning and answering system adopting initiative study-based questioning and answering method | |
| CN106844571B (en) | Method and device for identifying synonyms and computing equipment | |
| CN109492081B (en) | Text information searching and information interaction method, device, equipment and storage medium | |
| CN109710732B (en) | Information query method, device, storage medium and electronic equipment | |
| CN107679070B (en) | Intelligent reading recommendation method and device and electronic equipment | |
| CN105956053A (en) | Network information-based search method and apparatus | |
| CN109522397B (en) | Information processing method and device | |
| CN105354199A (en) | Scene information based entity meaning identification method and system | |
| CN108345612A (en) | A kind of question processing method and device, a kind of device for issue handling | |
| CN112612875A (en) | Method, device and equipment for automatically expanding query words and storage medium | |
| CN113326702A (en) | Semantic recognition method and device, electronic equipment and storage medium | |
| CN109492085B (en) | Answer determination method, device, terminal and storage medium based on data processing | |
| CN113886544A (en) | Text matching method, device, storage medium and computer equipment | |
| CN118115781A (en) | Label identification method, system, equipment and storage medium based on multi-mode model | |
| CN107562774A (en) | Generation method, system and the answering method and system of rare foreign languages word incorporation model | |
| CN113177061A (en) | Searching method and device and electronic equipment | |
| CN109684357B (en) | Information processing method and device, storage medium and terminal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20171031 |