CN109902161B - Answer processing method, device, equipment and storage medium of question-answering system - Google Patents
Answer processing method, device, equipment and storage medium of question-answering system Download PDFInfo
- Publication number
- CN109902161B CN109902161B CN201910102807.9A CN201910102807A CN109902161B CN 109902161 B CN109902161 B CN 109902161B CN 201910102807 A CN201910102807 A CN 201910102807A CN 109902161 B CN109902161 B CN 109902161B
- Authority
- CN
- China
- Prior art keywords
- attribute
- answer
- question
- entities
- answer subject
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The disclosure provides an answer processing method of a question-answering system, comprising the following steps: identifying a question subject entity of the question, and obtaining the asked attribute of the question and/or synonyms of the asked attribute; searching according to the knowledge graph to obtain m homonym answer subject entities corresponding to the question subject entities, wherein m is more than 1; and selecting the answer subject entity from the m homonym answer subject entities according to the heat distribution of the m homonym answer subject entities. The disclosure also provides an answer processing device, an electronic device and a computer readable storage medium of the question answering system.
Description
Technical Field
The present disclosure relates to an answer processing method of a question-answering system, an answer processing device of a question-answering system, an electronic device, and a computer-readable storage medium.
Background
In existing question-answering systems, for questions for which there is an ambiguous subject, the question-answering system selects one entity (e.g., the highest-heat) from among a plurality of possible entities to answer as the subject of the question. Thus, when the question subject entity which the user wants to inquire cannot be judged according to the question and other information, the requirement of the user cannot be met.
Moreover, in the existing system, the situation that the answers with the same name should be output together cannot be covered, for example, a non-famous entity with the heat less than ten thousand is required to be removed, for example, a different type of entity is not required to be triggered, etc.
Disclosure of Invention
In order to solve at least one of the above technical problems, the present disclosure provides an answer processing method of a question-answering system, an answer processing device of a question-answering system, an electronic device, and a readable storage medium.
According to one aspect of the disclosure, an answer processing method of a question-answering system includes: identifying a question subject entity of the question, and obtaining the asked attribute of the question and/or synonyms of the asked attribute; searching according to the knowledge graph to obtain m homonym answer subject entities corresponding to the question subject entities, wherein m is more than 1; and selecting the answer subject entity from the m homonym answer subject entities according to the heat distribution of the m homonym answer subject entities.
According to at least one embodiment of the present disclosure, when an answer subject entity is selected from m homonym answer subject entities according to a heat distribution, when the heat of the answer subject entity with the highest heat of the m homonym answer subject entities is low, a tolerance value of a heat difference between the answer subject entity with the highest heat and the rest of answer subject entities in the m homonym answer subject entities is large; and when the heat degree of the answer subject entity with the highest heat degree in the m same-name answer subject entities is high, the tolerance value of the heat degree difference between the answer subject entity with the highest heat degree and the rest answer subject entities in the m same-name answer subject entities is small.
According to at least one embodiment of the present disclosure, after selecting the answer subject entity from m homonym answer subject entities, the method further includes: the selected answer subject entities are distinguished.
In accordance with at least one embodiment of the present disclosure, differentiating the selected answer subject entities includes: solving the attribute intersection of the selected answer subject entities; deleting the attribute with the same attribute value in the subordinate intersection; and selecting a distinguishing attribute from the attribute intersection of the same attribute as the deletion attribute value to distinguish the same-name answer subject by using the distinguishing attribute.
According to at least one embodiment of the present disclosure, when the attribute intersection of the selected answer subject entities is solved, if there is no identical attribute having different attribute values between the answer subject entities, deleting the answer subject entity having the lowest heat among the selected answer subject entities, and solving the attribute intersection of the selected answer subject entities after deleting the answer subject entity having the lowest heat again.
According to at least one embodiment of the present disclosure, when selecting a distinguishing attribute, when an attribute of an attribute intersection is given priority, the distinguishing attribute is selected according to the priority of the attribute, and when an attribute of the attribute intersection is not given priority, the distinguishing attribute is selected according to the text similarity between attribute values of the attribute intersection.
According to at least one embodiment of the present disclosure, text similarity is determined by calculating an edit distance between attribute values of an attribute intersection, and an attribute of an attribute value having the smallest text similarity is selected as a distinguishing attribute.
According to another aspect of the present disclosure, an answer processing device of a question-answering system includes: the acquisition module is used for identifying a question subject entity of the question and obtaining the question attribute and/or synonym of the question attribute; the retrieval module is used for retrieving according to the knowledge graph to obtain m homonym answer subject entities corresponding to the question subject entities, wherein m is more than 1; and the selection module is used for selecting the answer subject entity from the m homonym answer subject entities according to the heat distribution of the m homonym answer subject entities.
According to yet another aspect of the present disclosure, an electronic device includes: a memory storing executable instructions; and a processor executing the executable instructions stored in the memory, causing the processor to perform the method described above.
According to yet another aspect of the disclosure, a readable storage medium has stored therein executable instructions which when executed by a processor are adapted to carry out the method described above.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the disclosure and together with the description serve to explain the principles of the disclosure.
Fig. 1 is a schematic flow chart of an answer processing method of a question-answering system according to one embodiment of the present disclosure.
Fig. 2 is a schematic flow chart of an answer processing method of the question-answering system according to one embodiment of the present disclosure.
Fig. 3 is a schematic flow chart of distinguishing steps of an answer processing method of a question-answering system according to one embodiment of the present disclosure.
Fig. 4 is a schematic flow chart of an answer processing device of the question-answering system according to one embodiment of the present disclosure.
Fig. 5 is a schematic view of an electronic device according to one embodiment of the present disclosure.
Detailed Description
The present disclosure is described in further detail below with reference to the drawings and the embodiments. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant content and not limiting of the present disclosure. It should be further noted that, for convenience of description, only a portion relevant to the present disclosure is shown in the drawings.
In addition, embodiments of the present disclosure and features of the embodiments may be combined with each other without conflict. The present disclosure will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
A question and answer system (Question Answering System, abbreviated as QA system) is used to answer various questions input or presented by the user. For example, the user asks "who is somehow wife" and the QA system answers "somehow kun"; the user asks "how much people are in china", and the QA system answers "13 billion".
Knowledge Graph (KG) is aimed at covering all entities in the world and information of the entities. The general representation method of the information of the entity is in a triplet form, such as: sometimes, wife, sometimes.
Question subject entity: in the questions received by the QA system, the question subject entity refers to the entity that is queried in the question and is the subject in the question. Such as: "Wednesday" in "who is the wife" is the subject entity of the question, and "China" in "how much of the population in China" is the subject entity of the question.
Homonym answer subject entity: in a question received by the QA system, a question subject entity may correspond to multiple homonym answer subject entities, and it is not possible to determine which answer subject entity the question asks by the question. For example, when a user asks "apple," it may refer to the answer subject entity "apple" as fruit, or it may refer to the answer subject entity "apple" as apple company. Thus, there will likely be two identical-name answer subject entities, one being an apple as fruit and the other being an apple as company. For another example, when the user asks "who the author of the orchid's sequence is," the question subject entity in the question is "orchid's sequence", and there are a plurality of answer subject entities of the orchid's sequence, for example, the orchid's sequence may refer to "orchid's sequence" of the written calligraphy work of Wangzhi; it may also refer to a song "blue pavilion" that is sung somewhere around, so that there will be two identical answer subjects, "blue pavilion of article" and "blue pavilion of song".
According to one embodiment of the present disclosure, an answer processing method of a question-answering system is provided.
As shown in fig. 1, the answer processing method 10 includes: step S11, identifying a question subject entity of the question, and obtaining the question attribute and/or synonym of the question attribute; step S12, searching according to the knowledge graph to obtain m homonym answer subject entities corresponding to the question subject entities, wherein m is more than 1; and step S13, selecting the answer subject entity from the m homonym answer subject entities according to the heat distribution of the m homonym answer subject entities.
In step S11, a question subject entity of the question is identified from the question of the user, the question attribute of the question is parsed, and synonyms of the question attribute are expanded.
By way of example, the question asked by the user is "who the orchid pavilion wrote? The "blue pavilion" is the main question entity, the "author" is the question attribute of the question, and the "author" is the synonym of the attribute "author" is expanded, for example, "word" is expanded (here, only an example, more synonyms can be expanded according to the situation), and the "blue pavilion" is the main question entity of the question, and the "author" and the "word" are the attributes of the question.
In step S12, searching is performed according to the knowledge graph to obtain m identical-name answer subject entities corresponding to the question subject entities, where m is greater than 1.
As one example, those entities that hit aliases may be used when retrieving in the knowledge-graph. For example, when the user asks "how big the sun is? By the term "", since the artist in korea is also suny in some alias, although his entity name in the knowledge graph is not suny, the artist is also hit according to the alias. This may be taken as one of a plurality of answer subject entities. As another example, for example, when the user asks "when to rewind the song? ", assuming that there is in the knowledge graph: the method comprises the steps of rewinding a sound, a Chua sound and a sound, wherein three homonymous answer subject entities, namely the sound, the sound and the sound, can be searched in a knowledge graph according to a subject entity 'rewinding', and the knowledge graph can be searched according to the subject entity 'rewinding'. The number "three" is here for illustrative purposes only, and in practice more may be present. As another example, if the user asks "when to rewind the song? The knowledge map is searched to obtain the rewinding of the singing around the singing, the rewinding of the singing of the Chua, and the rewinding of the singing of the Xiao. But there is no attribute of "release time" for "rewind of a singing somewhere around" because it can be culled out of the way as an option. The "rewind of Cai's singing" and "rewind of Xiao's singing" both have the attribute of "release time", and both are taken as the selected items. Again, an example is provided with the user's question "who is the blue pavilion written? By way of example, by searching in the knowledge graph, if it is assumed that the author of the lankton chapter is Wangzhi and the word of the lankton song is a party in the knowledge graph, then two homonymous answer subject entities, "lankton as article" and "lankton as song" can be searched in the knowledge graph.
In step S13, in the function of outputting the same-name entities together, the same-name answer subject entity that is more likely to be the user query is selected based on the heat distribution of the plurality of same-name answer subject entities, instead of just the entity with the highest heat.
In one embodiment of the present disclosure, when an answer subject entity is selected from m homonym answer subject entities according to a heat distribution, when the heat of the answer subject entity with the highest heat of the m homonym answer subject entities is low, a tolerance value of a heat difference between the answer subject entity with the highest heat and the rest of answer subject entities in the m homonym answer subject entities is large; and when the heat degree of the answer subject entity with the highest heat degree in the m same-name answer subject entities is high, the tolerance value of the heat degree difference between the answer subject entity with the highest heat degree and the rest answer subject entities in the m same-name answer subject entities is small.
The selection of answer subject entities based on heat distribution is described below in connection with specific embodiments.
In the question-answering system, when the answer with highest heat is searched according to the knowledge graph, the search is not stopped at this time, but the search is continued to judge whether the rest of the homonym answer subject entities contain attributes capable of answering the user questions.
And stopping searching when the heat difference between the rest homonym answer subject entities and the answer subject entity with the highest heat is overlarge. For example, the subject "china" is the same name answer subject entity, which has both "the people's republic of China", abbreviated as china, and has the chinese in "reed-source china", which refers to the world between people in japan myth, and possibly also "henna loyang", zhou Chaoshi loyang was called china. Among these entities, the heat of the people's republic of China is far higher than that of the other two entities, and the questioner really only concerns about the related attributes under the people's republic of China, so that the questioning and answering system does not need to search the latter two entities and directly returns the answers under the people's republic of China to meet the needs of users. For another example, when the problem of the user is "how big" and the highest heat is the one that is the actor, if there are multiple blobs (e.g., multiple persons of the same name) in the knowledge graph, the remaining persons of the same name are only non-famous persons of the same name (the heat is low) except for the actor blobs, and then the other persons can be rejected.
The heat gap "too large" described above can be fitted to the actual situation by a piecewise function. For example, when the hotness of the answer subject entities is low, for example, in the case where the hotness is only one, ten or hundred digits, the hotness of these entities is a large multiple of the difference in value, but the hotness is low for a real life person although the difference is large, so these entities are not much different in the hotness, so the tolerance value for the hotness difference of these entities is large, and when expressed by a piecewise function, for example, they can be regarded as no difference. As the heat of the entities increases, the tolerance for the gap between the heat of the entities becomes smaller. For example, taking the example that two entities exist, when the heat of the two entities is low, one is a single digit and the other is a hundred digits, although the difference between the two is about a hundred times, the two entities are regarded as no difference in heat because the heat is low and the tolerance value is large (for example, the difference between about a hundred times is not regarded as the difference is too large), and the two entities may be output together; however, when one of the two entities has a high heat, for example, a heat of ten thousand digits or more, and the other entity has a low heat, for example, a hundred digits, the difference between the two is about hundred times, and because the tolerance value is smaller, the difference is regarded as the difference "too large", so that only the entity having a high heat may be input.
According to another embodiment of the present disclosure, an answer processing method of a question-answering system is provided. As shown in fig. 2, the answer processing method 20 includes step S24 to distinguish selected answer subject entities, in addition to steps S21-S23, which are the same as steps S11-S13 of the method 10, respectively.
In step S24, the plurality of answer subject entities are distinguished according to the distinguishing attribute of the selected plurality of answer subject entities. In this step, since a number of possible answer subject entities have been found, a distinction is required between these entities, otherwise it cannot be determined which answer(s) are the answer(s) for the intended answer. In the present disclosure, the distinguishing attribute of the plurality of same-name answer subject entities is used for distinguishing, for example, the distinguishing attribute is found, thereby distinguishing the plurality of same-name answer subject entities. For example, both sun as a celestial body and sun as a person's external number have an age attribute, and when the user is informed that the age of a certain sun is several tens of years, the user can obviously know that the sun is a person with an external number of sun, not sun as a celestial body.
As shown in fig. 3, step S24 further includes: s241: solving the attribute intersection of the selected answer subject entities; s242: deleting the attribute with the same attribute value in the subordinate intersection; s243: and selecting distinguishing attributes from the attribute intersection sets of the attributes with the same deleting attribute value, so as to distinguish the same-name answer subjects by using the distinguishing attributes.
Specifically, in step S241, a plurality of answer subject entities are compared to find intersections of their attributes. When the attribute intersection is obtained, if no attribute which is owned by all but has different attribute values exists among the homonym answer subject entities except the answer attribute, the entity with the lowest heat is removed, the intersection is obtained again for the rest answer subject entities, and one or more attributes which are owned by all the current entities and have different attribute values are always found.
From this intersection set, the attributes having the same attribute values may be deleted in step S242, thereby finding out those attributes having the distinction degree, that is, the respective attribute values of the plurality of homonym answer subject entities are all different. The attributes in this intersection set can all distinguish between these answer subject entities because each answer subject entity possesses these attributes and the respective attribute values are different.
In step S243, when the attribute of the attribute intersection is given priority, the discriminating attribute is selected according to the priority of the attribute, and when the attribute of the attribute intersection is not given priority, the discriminating attribute is selected according to the text similarity between the attribute values of the attribute intersection. Specifically, when an attribute is given priority in advance, then the distinguishing attribute may be selected directly in accordance with the priority given in advance. In the present disclosure, some/one attribute may be given a higher priority, while other attributes have a lower priority, so that the attribute with the highest priority may be used as the distinguishing attribute. Preferably, a distinguishing attribute blacklist may also be provided, for example, some attributes with distinguishing degrees are not used as distinguishing attributes of distinguishing entities. For example, if the user asks the host "thank somewhere" to mention "na", but the name "na" is a lot of people, it is assumed that the attributes of these "na" can be distinguished, including birth place, height, age, full name. But the degree of distinction of the full name may be significantly higher than the place of birth, height, age. The full name may be used as a distinguishing attribute of choice. If other attributes are selected, for example, tell the user that one na is born in Sichuan and one na is born in Siam, then the user needs to have knowledge of the place of birth, and if not, it cannot be determined which one is the one of interest. If the full name is taken as the distinguishing attribute, the user can directly tell that na is about to be about, and the user can clearly judge that he is concerned about.
In another case, for example, it is difficult to assign priorities to the attributes one by one in advance due to the variety of the attributes, and when the attributes to which priorities have been assigned are not included in the attribute intersections between the entities, it is possible to select a discrimination attribute from among the attributes by calculating the text similarity between attribute values of the attributes in the attribute intersections, for example, selecting the attribute having the smallest text similarity as the discrimination attribute. Text similarity may be calculated, for example, by a cosine similarity algorithm. In the present disclosure, the edit distance (e.g., levenshtein distance) between the attribute values of these attributes may optionally be calculated to select a distinguishing attribute, and when the edit distance is larger, the text similarity is smaller, at which time the attribute having the largest edit distance may be selected as the distinguishing attribute.
As an example, there are three pieces of software named "trusted", for example, in which attribute values of attribute "software category" are "chat communication software", "communication mobile office software" and "communication/Android", respectively, in their attribute intersections; the attribute values of the attribute "developer" are "netease and telecommunications", "sitter odd" and "jinzheng" respectively. By calculating the edit distance between the attribute values of the respective attributes, the edit distance of the attribute value of the attribute "developer" is larger than the edit distance of the attribute value of the attribute "software category", and thus the attribute "developer" can be selected as the discrimination attribute.
According to an optional embodiment of the disclosure, the answer processing method of the disclosure further comprises a step of integrating the answer entities. At this time, the natural language is utilized to integrate the plurality of answers, and a complete utterance which is easy to understand by the user is integrated and returned to the user. For the attributes with pre-set priorities, a specific template can be adopted to enable the integrated sentences to be more smooth, and for the attributes without pre-set priorities, the integration can be carried out according to the names and the attribute values of the attributes, and the like.
As an example, when the selected answer subject entity is plural, for example, the plural answer subject entities and the related information are integrated, and the integrated content is handled as an answer. The output answers include: the distinguishing attribute of the selected answer subject entity, the corresponding attribute of the selected answer subject entity and the attribute value of the corresponding attribute may also include the number of the selected answer subject entities.
For example, for the question "who is the orchid motif? And retrieving from the knowledge graph to obtain two identical-name answer subject entities, namely 'the orchid pavilion of the article' and 'the orchid pavilion of the song'. The "song" and the "article" are distinguishing attributes capable of distinguishing two homonym answer subjects, the "composition" and the "author" are corresponding attributes of two homonym answer subjects, and the "party" and the "Wangzhi" are attribute values of two corresponding attributes respectively. The information obtained may be integrated at this time to give an answer, for example, the answer "the word in the orchid order as a song is somebody else, and the author in the orchid order as an article is Wang".
For example, for the question "when to rewind the song? The 'round-the-hole' and the 'Chua' of the 'round-the-hole' of the 'Zhi' of the 'round-the' hole 'of the' Xiao 'of the' Zhi 'of the' are retrieved from the knowledge graph, and the 'round-the-hole' of the 'round-the' hole 'is not issued with the time attribute, so that only the' Chua 'of the' round-the 'hole' and the 'Xiao' of the 'hole' are selected. At this time, the distinguishing attribute "singer" of the two same-name answer subject entities is used as the distinguishing attribute for distinguishing. The release time of the rewinding tape which can generate a certain singing of the Chua is 2 months 2004; the release time of the rewinding of a singing was 11 months in 2009.
In the present disclosure, preferably, an answer may be provided to a user according to the following framework. It is to be understood that this frame is for illustrative purposes only and that one skilled in the art may use a suitable frame depending on the actual situation.
Taking the number of selected answer subject entities as two as an example, continuing with the question "who was the blue pavilion written? By way of example of "the framework may be" I know AAA has BBB's, DDD of AAA of CCC (as/name) is EEE, GGG of AAA of FFF (as/name) is HH, … … ". The answer processing according to this framework was "there are 2 in the orchid order i know, the word of the orchid order as a song is somebody, and the author of the orchid order as an article is Wang".
In this framework, "AAA" represents a question subject entity such as "Lantern" or "Lantern" for example. "BBB" means the number of resulting multiple homonym answer subject entities, such as "2". The AAA of the CCC and the AAA of the FFF represent a plurality of homonym answer subjects entities such as "the orchid sequence of songs" and "the orchid sequence of articles", wherein the CCC and the FFF represent distinguishing attributes such as "songs" and "articles" of the homonym answer subjects entities, which can distinguish the homonym answer subjects entities, and further, for example, taking a song "rewind" as an example, there may be a singer distinction (distinguishing attribute) at this time, namely "a singing of a cai" and "a singing of a xiao. "DDD" and "GGG" represent the corresponding attributes of a plurality of homonym answer subject entities, in the case of the lanktree, "DDD" being the corresponding attribute "word" and "GGG" being the corresponding attribute "author", where the two corresponding attributes are different. However, the two corresponding attributes may be the same, for example, in the case of rewinding songs, "DDD" and "GGG" are both corresponding attributes "release time". The "EEE" and "HHH" represent attribute values of the corresponding attributes of the same-name answer subject entities, for example, the attribute value of "word of the orchid sequence as a song" is "party" and the attribute value of "author of the orchid sequence as an article" is "Wang", the attribute value of "release time of the rewind of a chai" is "2004 month 2", and the attribute value of "release time of the rewind of a shou" is "2009 month 11".
The words "as/fame" in brackets in the above-described framework depend on the attribute values of what attribute "CCC" and "FFF" are "AAA", and are used to indicate that "CCC" and "FFF" are attributes of "AAA".
Finally, the text obtained is output as an answer, which can then be converted to speech for output according to techniques commonly used in the art.
The method can improve the accuracy of the question-answering system, so that the experience of the user is improved by providing more information for the user.
According to still another embodiment of the present disclosure, an answer processing device of a question-answering system is provided. As shown in fig. 4, the answer processing device 40 of the question-answering system includes an acquisition module 41, a retrieval module 42, and a selection module 43.
The obtaining module 41 identifies the question subject entity of the question, and obtains the question attribute and/or the synonym of the question attribute; the retrieval module 42 retrieves according to the knowledge graph to obtain m identical-name answer subject entities corresponding to the question subject entities, wherein m is more than 1; and a selection module 43 for selecting the answer subject entity from the m homonym answer subject entities according to the heat distribution of the m homonym answer subject entities.
And the processing procedures executed in the above modules respectively correspond to the corresponding procedures specifically described in the above method.
As shown in fig. 4, the answer processing device 40 of the question-answering system may further include a differentiating module 44 and an integrating module 45. The differentiating module 44 is used for differentiating the selected answer subject entities, and the integrating module 45 is used for integrating the answer entities. The processing procedures executed by the two modules respectively correspond to the distinguishing step and the integrating step of the method.
The present disclosure also provides an electronic device, as shown in fig. 5, including: a communication interface 1000, a memory 2000 and a processor 3000. The communication interface 1000 is used for communicating with external devices for data interactive transmission. A computer program executable on the processor 3000 is stored in the memory 2000. The processor 3000 implements the method in the above embodiment when executing the computer program. The number of the memories 2000 and the processors 3000 may be one or more.
The memory 2000 may include a high-speed RAM memory or may further include a non-volatile memory (non-volatile memory), such as at least one magnetic disk memory.
If the communication interface 1000, the memory 2000 and the processor 3000 are implemented independently, the communication interface 1000, the memory 2000 and the processor 3000 may be connected to each other through a bus and perform communication with each other. The bus may be an industry standard architecture (ISA, industry Standard Architecture) bus, an external device interconnect (PCI, peripheralComponent) bus, or an extended industry standard architecture (EISA, extended Industry StandardComponent) bus, among others. The buses may be classified as address buses, data buses, control buses, etc. For ease of illustration, only one thick line is shown in the figure, but not only one bus or one type of bus.
Alternatively, in a specific implementation, if the communication interface 1000, the memory 2000, and the processor 3000 are integrated on a chip, the communication interface 1000, the memory 2000, and the processor 3000 may perform communication with each other through internal interfaces.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and further implementations are included within the scope of the preferred embodiment of the present disclosure in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the embodiments of the present disclosure. The processor performs the various methods and processes described above. For example, method embodiments in the present disclosure may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as a memory. In some embodiments, part or all of the computer software program may be loaded and/or installed via memory and/or a communication interface. One or more of the steps of the methods described above may be performed when a computer software program is loaded into memory and executed by a processor. Alternatively, in other embodiments, the processor may be configured to perform one of the methods described above in any other suitable manner (e.g., by means of firmware).
Logic and/or steps represented in the flowcharts or otherwise described herein may be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions.
For the purposes of this description, a "computer-readable storage medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It should be understood that portions of the present disclosure may be implemented in hardware, software, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Those of ordinary skill in the art will appreciate that all or a portion of the steps implementing the method of the above embodiments may be implemented by a program to instruct related hardware, and the program may be stored in a computer readable storage medium, where the program includes one or a combination of the steps of the method embodiments when executed.
Furthermore, each functional unit in each embodiment of the present disclosure may be integrated into one processing module, or each unit may exist alone physically, or two or more units may be integrated into one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented in the form of software functional modules and sold or used as a stand-alone product. The storage medium may be a read-only memory, a magnetic or optical disk, or the like.
In the description of the present specification, reference to the terms "one embodiment/mode," "some embodiments/modes," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment/mode or example is included in at least one embodiment/mode or example of the present application. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment/manner or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments/modes or examples. Furthermore, the various embodiments/implementations or examples described in this specification and the features of the various embodiments/implementations or examples may be combined and combined by persons skilled in the art without contradiction.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present application, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
It will be appreciated by those skilled in the art that the above-described embodiments are merely for clarity of illustration of the disclosure, and are not intended to limit the scope of the disclosure. Other variations or modifications will be apparent to persons skilled in the art from the foregoing disclosure, and such variations or modifications are intended to be within the scope of the present disclosure.
Claims (7)
1. An answer processing method of a question-answering system, comprising:
identifying a question subject entity of a question, and obtaining a question attribute and/or a synonym of the question attribute of the question;
searching according to the knowledge graph to obtain m homonym answer subject entities corresponding to the question subject entities, wherein m is more than 1; and
selecting answer subject entities from the m homonym answer subject entities according to the heat distribution of the m homonym answer subject entities, and distinguishing the selected answer subject entities;
the distinguishing the selected answer subject entity comprises:
solving the attribute intersection of the selected answer subject entities;
deleting the attribute with the same attribute value from the attribute intersection; and
selecting distinguishing attributes from attribute intersections of the same attribute with the deleted attribute values so as to distinguish the same-name answer subjects by using the distinguishing attributes;
the selecting the distinguishing attribute from the attribute intersection of the same attribute of the deleted attribute value comprises:
selecting a discriminating attribute according to the priority of the attribute when the attribute of the attribute intersection is given priority;
when the attributes of the attribute intersection are not given priority, a discriminating attribute is selected according to the text similarity between the attribute values of the attribute intersection.
2. The method of claim 1, wherein when an answer subject entity is selected from m homonym answer subject entities based on the heat distribution,
when the heat degree of the answer subject entity with the highest heat degree in the m same-name answer subject entities is low, the tolerance value of the heat degree difference between the answer subject entity with the highest heat degree and the rest answer subject entities in the m same-name answer subject entities is large; and
when the heat degree of the answer subject entity with the highest heat degree of the m same name answer subject entities is high, the heat degree is highest
The tolerance value of the heat difference between the answer subject entity and the rest of the m homonym answer subject entities is small.
3. The method of claim 1, wherein when the attribute intersection of the selected answer subject entities is found, if there is no identical attribute having a different attribute value between the answer subject entities, then deleting the answer subject entity having the lowest heat among the selected answer subject entities, and finding the attribute intersection of the selected answer subject entities after the answer subject entity having the lowest heat is deleted again.
4. The method of claim 1, wherein the text similarity is determined by calculating an edit distance between attribute values of the attribute intersection, and an attribute of an attribute value having a smallest text similarity is selected as the distinguishing attribute.
5. An answer processing device of a question-answering system, comprising:
the acquisition module is used for identifying a question subject entity of a question and obtaining a question attribute and/or a synonym of the question attribute of the question;
the retrieval module is used for retrieving according to the knowledge graph to obtain m homonym answer subject entities corresponding to the question subject entities, wherein m is more than 1; and
the selection module is used for selecting answer subject entities from the m homonym answer subject entities according to the heat distribution of the m homonym answer subject entities and distinguishing the selected answer subject entities;
the selection module is used for distinguishing the selected answer subject entities according to the answer subjects entities and is also used for:
solving the attribute intersection of the selected answer subject entities;
deleting the attribute with the same attribute value from the attribute intersection; and
selecting distinguishing attributes from attribute intersections of the same attribute with the deleted attribute values so as to distinguish the same-name answer subjects by using the distinguishing attributes;
the selecting the distinguishing attribute from the attribute intersection of the same attribute of the deleted attribute value comprises:
selecting a discriminating attribute according to the priority of the attribute when the attribute of the attribute intersection is given priority;
when the attributes of the attribute intersection are not given priority, a discriminating attribute is selected according to the text similarity between the attribute values of the attribute intersection.
6. An electronic device, comprising:
a memory storing executable instructions; and
a processor executing executable instructions stored by the memory, causing the processor to perform the method of any one of claims 1 to 4.
7. A readable storage medium having stored therein executable instructions which when executed by a processor are adapted to carry out the method of any one of claims 1 to 4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910102807.9A CN109902161B (en) | 2019-02-01 | 2019-02-01 | Answer processing method, device, equipment and storage medium of question-answering system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910102807.9A CN109902161B (en) | 2019-02-01 | 2019-02-01 | Answer processing method, device, equipment and storage medium of question-answering system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109902161A CN109902161A (en) | 2019-06-18 |
CN109902161B true CN109902161B (en) | 2023-10-20 |
Family
ID=66944621
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910102807.9A Active CN109902161B (en) | 2019-02-01 | 2019-02-01 | Answer processing method, device, equipment and storage medium of question-answering system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109902161B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111309863B (en) * | 2020-02-10 | 2020-12-18 | 北京声智科技有限公司 | Natural language question-answering method and device based on knowledge graph |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106844603A (en) * | 2017-01-16 | 2017-06-13 | 竹间智能科技(上海)有限公司 | The computational methods and device, application process and device of entity hot topic degree |
CN106919655A (en) * | 2017-01-24 | 2017-07-04 | 网易(杭州)网络有限公司 | A kind of answer provides method and apparatus |
CN107247736A (en) * | 2017-05-08 | 2017-10-13 | 广州索答信息科技有限公司 | The kitchen field answering method and system of a kind of knowledge based collection of illustrative plates |
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 |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11182681B2 (en) * | 2017-03-15 | 2021-11-23 | International Business Machines Corporation | Generating natural language answers automatically |
-
2019
- 2019-02-01 CN CN201910102807.9A patent/CN109902161B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106844603A (en) * | 2017-01-16 | 2017-06-13 | 竹间智能科技(上海)有限公司 | The computational methods and device, application process and device of entity hot topic degree |
CN106919655A (en) * | 2017-01-24 | 2017-07-04 | 网易(杭州)网络有限公司 | A kind of answer provides method and apparatus |
CN107247736A (en) * | 2017-05-08 | 2017-10-13 | 广州索答信息科技有限公司 | The kitchen field answering method and system of a kind of knowledge based collection of illustrative plates |
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 |
Also Published As
Publication number | Publication date |
---|---|
CN109902161A (en) | 2019-06-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4778434B2 (en) | Database query user interface | |
US20120215806A1 (en) | Navigation system data base system | |
CN101819578A (en) | Retrieval method, method and device for establishing index and retrieval system | |
CN108304375A (en) | A kind of information identifying method and its equipment, storage medium, terminal | |
CN109597874B (en) | Information recommendation method, device and server | |
JP2009217585A (en) | Information-retrieving device, information-retrieving system, and information-retrieving method | |
JPH06266780A (en) | Character string retrieving method by semantic pattern recognition and device therefor | |
CN104699784A (en) | Data searching method and device based on interactive input | |
JP2009522695A (en) | How to attach metadata to documents and document objects using the operating system user interface | |
CN104469029A (en) | Method and device for telephone number query through voice | |
CN109902161B (en) | Answer processing method, device, equipment and storage medium of question-answering system | |
CN109753557B (en) | Answer output method, device, equipment and storage medium of question-answering system | |
JP2011013705A (en) | Name searching apparatus | |
US8793271B2 (en) | Searching documents using a dynamically defined ignore string | |
CN109815320B (en) | Answer generation method, device, equipment and storage medium of question-answering system | |
CN112836021B (en) | Intelligent search system of library | |
CN112989011B (en) | Data query method, data query device and electronic equipment | |
CN112988986B (en) | Man-machine interaction method, device and equipment | |
CN114265846A (en) | Data operation method and device, electronic equipment and storage medium | |
CN112069273A (en) | Address text classification method and device, electronic equipment and storage medium | |
CN101366030A (en) | Structured data storage | |
JP2011076194A (en) | System, method and program for creating topic concrete expression dictionary | |
WO2005072359A2 (en) | Method and apparatus for determining when a user has ceased inputting data | |
CN113158607B (en) | SystemVerilog overlay set generation method and device | |
CN112861021B (en) | Multi-point-based house source searching method and system |
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 | ||
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20211117 Address after: 210034 floor 8, building D11, Hongfeng Science Park, Nanjing Economic and Technological Development Zone, Jiangsu Province Applicant after: New Technology Co.,Ltd. Applicant after: VOLKSWAGEN (CHINA) INVESTMENT Co.,Ltd. Address before: 100094 1001, 10th floor, office building a, 19 Zhongguancun Street, Haidian District, Beijing Applicant before: MOBVOI INFORMATION TECHNOLOGY Co.,Ltd. |
|
GR01 | Patent grant | ||
GR01 | Patent grant |