WO2018227930A1

WO2018227930A1 - Method and device for intelligently prompting answers

Info

Publication number: WO2018227930A1
Application number: PCT/CN2017/118746
Authority: WO
Inventors: 王经委; 张杰伟; 张霄; 贺坚; 程涛远
Original assignee: 百度在线网络技术（北京）有限公司
Priority date: 2017-06-15
Filing date: 2017-12-26
Publication date: 2018-12-20
Also published as: CN107436916A; CN107436916B

Abstract

A method and device for intelligently prompting answers. The method comprises: analyzing and extracting information features of a question Q, the information features comprising lexical features S2 (Q), syntactic features S4 (Q), semantic lexical features S3 (Q), and semantic syntactic features S5 (Q) (S10); computing the similarity of each type of information features between the question Q and questions Qi in a question bank, and obtaining similarity Sim (Q, Qi) between the question Q and the questions Qi in the question bank, wherein the question bank contains the questions Qi and corresponding answers (S20); and sorting the questions Qi in the question bank according to similarity Sim (Q, Qi), and providing the answers to some questions Qi in the question bank which are among the most front of the sorted results to a user (S30). According to the method, intelligent and accurate analyses for the questions can be achieved, the corresponding answers to the questions are quickly screened out and obtained, and therefore the response speed and accuracy for answering the questions are greatly improved.

Description

Method and device for intelligent prompt answer

Cross-reference to related applications

This application claims the priority of the Chinese patent application number "201710452312.X" submitted by Baidu Online Network Technology (Beijing) Co., Ltd. on June 15, 2017, entitled "Method and Apparatus for Answering Intelligent Tips".

Technical field

The present disclosure generally relates to the field of computer technologies, and in particular, to a method and apparatus for intelligently prompting answers.

Background technique

With the rise of e-commerce platforms, more and more industries are beginning to combine with e-commerce. Compared with traditional sales channels, face-to-face direct communication, how to communicate effectively and professionally with customers online, is very important to enhance the company's online sales.

In the customer service system, the customer and the company's customer service communicate through web pages, communication software, and telephone. For example, the customer asks questions about the company's products or business, and the customer service gives answers to customer questions. However, due to the limitations and differences of customer service personnel's business knowledge, it is difficult to guarantee the accuracy and consistency of the answers given. In order to make up for the lack of customer service personnel in the business knowledge, the company needs to conduct regular business training, and the cost is high.

Some systems are known to add a quick reply function to the above problems. With this function, the customer service can set common answers in the system in advance, and when the customer asks the same question, they can choose the pre-set answer. The drawback of this method is that it needs the customer service to judge which answer the customer question is matched with. When the preset number of answers is small, it is difficult to meet the demand. When the preset number of answers is large, the customer service is difficult to quickly find the matching answer and cannot respond quickly. Therefore, existing systems and methods can only assist with small, simple problems.

Summary of the invention

In view of the above-mentioned drawbacks or deficiencies in the prior art, it is desirable to provide an efficient, flexible and intelligent method of prompting answers.

In a first aspect, an embodiment of the present application provides a method for intelligently prompting an answer, including:

An information feature of the problem Q is analyzed and extracted, the information feature including a lexical feature S ₂ (Q), a syntactic feature S ₄ (Q), a semantic lexical feature S ₃ (Q), and a semantic syntactic feature S ₅ (Q);

Wherein each class information is calculated between Q and Q of the problem _I library have problems similarity, similarity Sim is obtained and the problem Q Q _I library have the problem of (Q, Q _I), wherein the library have Contains the question Q _i and the corresponding answer;

The problem library problem Q _i is sorted according to the similarity Sim(Q, Q _i ), and the answers of several question database questions Q _i with the highest ranking result are provided to the user.

In some embodiments, the method for intelligently prompting an answer includes: the problem library includes a public problem library and a private problem library; and the similarity of the questions in the question and question library Sim(Q, Q _i ) is prioritized by the high priority and the private question. The principles are sorted; in response to the final answer determined by the user based on the provided answer; the question Q and the final answer are combined and added to the private question base.

In a second aspect, the embodiment of the present application further provides an apparatus for intelligently prompting an answer, including:

A feature extraction unit configured to analyze and extract an information feature of the question Q, the information feature including a lexical feature S ₂ (Q), a syntax feature S ₄ (Q), a semantic lexical feature S ₃ (Q), and a semantic syntax feature S ₅ (Q);

Similarity similarity calculation unit, wherein the configuration information for each class is calculated between Q and the problem library have problems of Q _I, Q of the problem is obtained and the similarity Sim library have the problem of _I Q (Q, Q _i ), where the question library contains the question Q _i and the corresponding answer;

The answer screening unit is configured to sort the problem library problem Q _i according to the similarity Sim(Q, Q _i ), and provide the answer of the plurality of problem library questions Q _i with the highest ranking result to the user.

In a third aspect, the embodiment of the present application further provides an apparatus, including:

One or more processors and memories;

Therein, the memory includes instructions executable by the one or more processors to cause the one or more processors to perform the method of intelligent prompt answers provided in accordance with various embodiments of the present application.

In a fourth aspect, an embodiment of the present application further provides a computer readable storage medium storing a computer program, the computer program causing a computer to perform a method for answering an intelligent prompt according to embodiments of the present application.

The method for intelligently prompting answers provided by the embodiments of the present application can obtain the similarity between the problem and the problem database problem by calculating the similarity degree of each type of information feature between the problem and the problem library problem, and can realize intelligent and accurate analysis of the problem; Sorting similarity helps to quickly filter the answers to the questions, so that the customer service can quickly and accurately answer customer questions, greatly improving the response speed of the customer service and the accuracy of the response;

The method for intelligent prompt answer provided by some embodiments of the present application further ensures the accuracy and consistency of the answer by configuring the public problem database and the private question, and satisfies the flexibility and personalization of the answer, and effectively improves the service of the customer service. quality.

DRAWINGS

Other features, objects, and advantages of the present application will become more apparent from the detailed description of the accompanying drawings.

FIG. 1 illustrates an exemplary flow chart of a method of intelligent prompting an answer in accordance with an embodiment of the present application;

2 is a schematic structural diagram of an apparatus for answering an intelligent prompt provided by an embodiment of the present application;

FIG. 3 shows a schematic structural view of a device suitable for implementing an embodiment of the present application.

detailed description

The present application will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention, rather than the invention. It should also be noted that, for the convenience of description, only parts related to the invention are shown in the drawings.

It should be noted that the embodiments in the present application and the features in the embodiments may be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings.

FIG. 1 illustrates an exemplary flow chart of a method of intelligently prompting an answer in accordance with an embodiment of the present application.

As shown in Figure 1, the method of intelligently prompting answers includes:

Step S10: analyzing and extracting information features of the question Q, the information features including lexical features S ₂ (Q), syntactic features S ₄ (Q), semantic lexical features S ₃ (Q), and semantic syntactic features S ₅ (Q).

Specifically, in this embodiment, step S10 may be, but is not limited to, implemented as follows:

Segmenting the problem Q to obtain the word segmentation sequence S ₁ (Q);

Performing part-of-speech tagging on the segmentation word sequence S ₁ (Q) to obtain the lexical feature S ₂ (Q);

Semantic annotation of the lexical feature S ₂ (Q) according to the semantic dictionary, obtaining the semantic lexical feature S ₃ (Q);

Performing adjacent combinations on the lexical feature S ₂ (Q) to obtain a syntactic feature S ₄ (Q);

The semantic lexical feature S ₃ (Q) is adjacently combined to obtain a semantic syntactic feature S ₅ (Q).

Preferably, the participle sequence S ₁ (Q) is subjected to part-of-speech tagging, and obtaining the lexical feature S ₂ (Q) can be obtained as follows:

To perform part-of-speech tagging on the segmentation sequence S ₁ (Q), and retain the nouns, verbs, adjectives and interrogative words, to obtain the part-of-speech tagging sequence;

The lexical feature S ₂ (Q) is the part of the part of speech tagging.

Preferably, the lexical feature S ₂ (Q) is semantically annotated according to the semantic dictionary, and the obtained semantic lexical feature S ₃ (Q) can be obtained as follows:

Finding a semantic dictionary, converting words registered in the semantic dictionary in the lexical feature S ₂ (Q) into corresponding meanings, and retaining words not registered in the semantic dictionary as meanings, thereby lexical features S ₂ (Q ) converted to a sequence of senses;

The semantic lexical feature S ₃ (Q) is the original sequence.

For example, for question Q "How to improve my keyword rankings", the analysis extraction process is as follows:

1. Question segmentation, get the corresponding word segmentation sequence S ₁ (Q): how / improve / me / / keyword / ranking;

2. part of speech tagging, and retain nouns, verbs, adjectives and interrogative words, get lexical features S ₂ (Q), that is, part of speech tagging sequence: how (question word) / improve (verb) / keyword (noun) / ranking ( verb);

3. Find the semantic dictionary, convert the words registered in the semantic dictionary in the lexical feature S ₂ (Q) into the corresponding semantics, and retain the words not registered in the semantic dictionary as the original, thereby the lexical feature S ₂ (Q) is converted to the original sequence, that is, the semantic lexical feature S ₃ (Q): Ka35B0 (how) / Ie12A0 (improvement) / keyword (keyword) / Dd07A0 (rank), where "keyword" is not registered in In the semantic dictionary, it is reserved as a Yiyuan;

4. morphological characteristics S ₂ (Q) for the adjacent combinations, syntactic features give S ₄ (Q): <how to improve> / <increased keyword> / <keyword ranking>;

5. Perform adjacent combination on the semantic lexical feature S ₃ (Q) to obtain the semantic syntax feature S ₅ (Q): <Ka35B0, Ie12A0>/<Ie12A0, keyword>/<keyword, Dd07A0>.

The above method for analyzing and extracting the feature information feature has four levels of lexical features S ₂ (Q), syntactic features S ₄ (Q), semantic lexical features S ₃ (Q) and semantic syntactic features S ₅ (Q) of the problem sentence. The information is analyzed and extracted to lay the foundation for the accuracy of subsequent similarity calculations.

In more embodiments, different methods may be used to analyze and extract the lexical features S ₂ (Q), syntactic features S ₄ (Q), semantic lexical features S ₃ (Q) and semantic syntactic features of the problem Q according to actual needs. S ₅ (Q) can achieve the same technical effect.

Step S20: the degree of similarity between information characterizing each class and the Q calculation library have problems Q _i, Q obtain similarity Sim problems and problems of libraries question _I Q (Q, Q _i), wherein the library have Contains the question Q _i and the corresponding answer.

Specifically, in this embodiment, the similarity Sim (S _j (Q), S _j (Q _i )) of each type of information feature between the problem Q and the problem Q _{i in the} problem library may be calculated as:

Based on the similarity of each type of information feature described above, the similarity Sim(Q,Q _i ) of the problem Q and the problem Q _i in the problem library can be calculated as follows:

_{Wherein, Com (S j (Q)} , S j (Q i)) is S _j (Q) and S _j (Q _i) of the common element _{number, Num (S j (Q)} , S j (Q i)) For the maximum number of elements in S _j (Q) and S _j (Q _i ), σ is a smoothing parameter, and the element is a word, a word combination, a sense or a combination of sense, j=2, 3, 4 or 5.

The smoothing parameter σ is configured according to experience and can be adjusted in the subsequent feedback based on the effect of the smart answer prompt. Set smoothing parameter σ, avoid _{Com (S j (Q),} S j (Q i)) and _{Num (S j (Q),} S j (Q i)) resulting in little difference similarity Sim (S _j (Q), S _j (Q _i )) calculation is inaccurate, effectively improving the discrimination between the two to improve the accuracy of the calculation.

In this embodiment, the similarity Sim(S _j (Q), S _j (Q _i )) of each type of information feature between the problem Q and the problem Q _{i in the} problem library is calculated by the above formula (1), and the lexical method is used. The information of the four levels of the feature S ₂ (Q), the syntactic feature S ₄ (Q), the semantic lexical feature S ₃ (Q) and the semantic syntactic feature S ₅ (Q) are combined, and the problem is calculated by the above formula (2). library have a problem Q _i of the similarity Sim (Q, Q _i), thus greatly improving the accuracy and reliability of Q and Q _i question bank similarity calculation.

In some embodiments, the similarity of question Q to question Q _i can be calculated for all questions Q _i in the problem library, followed by subsequent operations. Such an embodiment can traverse all of the questions Q _i in the problem library so as not to miss the appropriate answer. In other embodiments, the similarity of the problem Q to the partial problem may be calculated for only some of the questions in the problem library for subsequent processing. In such an embodiment, when the problem library is large, the candidate problem can be selected by preliminary screening (for example, keyword matching, etc.), thereby reducing the amount of calculation and further improving the response speed.

In more embodiments, the method provided by the present application is not limited to the calculation methods of the above formulas (1) and (2), and more different calculation methods may be configured to calculate between the problem Q and the problem Q _{i in the} problem library. The similarity of each type of information feature Sim(S _j (Q), S _j (Q _i )), the problem and the similarity of the problem in the problem library Sim(Q,Q _i ), as long as each type of information feature The similarity Sim (S _j (Q), S _j (Q _i )) obtains the similarity of the problem and the problem in the problem library Sim(Q, Q _i ), and the same technical effect can be achieved.

Step S30: Sorting the problem library problem Q _i according to the similarity Sim(Q, Q _i ), and providing the answer of the plurality of problem library questions Q _i with the highest ranking result to the user.

Particular problem, in the present embodiment, the problem according to the similarity Sim (Q, Q _i) of the library have problems Q _I sort order is the result of the most forward arsenals Q _I, which is the most similar to the problem of Q Therefore, its corresponding answer is provided to the user as the closest answer for reference.

In this embodiment, the sorting and screening based on the similarity Sim(Q, Q _i ) reduces the time for obtaining the prompt answer, and greatly improves the response speed of the reply.

Preferably, the problem library comprises a public problem library and a private problem library, wherein the public problem library is composed of questions and standard answers in the collection business scope of the company, and the private problem library is customized for each user according to their own needs.

Further preferably, the problem library problem Q _i is sorted according to the similarity high priority and private problem prioritization order, and some of the top results (for example, 3, 5, or 10) problem libraries are sorted. The answer to question Q _i is provided to the user.

Further preferably, based on the answers to the top question number questions Qi of the ranking result, the user determines the final answer as the answer to question Q; the question Q and the final answer are combined and added to the private question base.

In the above embodiment, the reasonable sorting and screening rules further help to obtain the optimal prompt answer quickly and accurately, thereby further improving the speed and accuracy of the question answer. The user (for example, customer service) can directly answer the customer's answer by selecting the appropriate answer from the provided prompt answer, or indirectly after answering the prompt answer. Further, by configuring the public problem database and the private problem database, the accuracy and consistency of the reply are ensured, and the flexibility and personalization of the reply are realized. Private problem libraries can be used only by users themselves or only by groups of users with permissions.

FIG. 2 is a schematic structural diagram of an apparatus for providing an intelligent prompt answer according to an embodiment of the present application. The device of the smart cue answer shown in FIG. 2 may correspond to any of the methods previously described in connection with FIG.

As shown in FIG. 2, in this embodiment, the present application provides an apparatus for intelligently prompting an answer, including:

The feature extraction unit 10 is configured to analyze and extract information features of the question Q, including lexical features S ₂ (Q), syntactic features S ₄ (Q), semantic lexical features S ₃ (Q), and semantic syntactic features S ₅ (Q);

Similarity calculating unit 20 is configured to feature information of the similarity between each type of calculation and the library have problems Q Q _I, Q and the problem is obtained similarity Sim library have the problem of _I Q (Q, Q _I ), where the problem library contains the question Q _i and the corresponding answer;

The answer screening unit 30, configured according to the similarity of the question database question Q _i sort, the sorting result answer forwardmost Problems arsenals Q _i provided to the user.

Specifically, in the present embodiment, the feature extraction unit 10 may be configured to analyze and extract the information features of the question Q as follows:

Segmenting the problem Q to obtain the word segmentation sequence S ₁ (Q);

Lexical wherein S ₂ (Q) for the part of speech tagging sequence.

The semantic lexical feature S ₃ (Q) is the original sequence.

For example, the feature extraction unit 10 analyzes the extraction process for the question Q "How to improve my keyword ranking":

4. Perform adjacent combination on the lexical feature S ₂ (Q) to obtain the syntactic feature S ₄ (Q): <how, improve>/<improvement, keyword>/<keyword, ranking>;

The method of analyzing and extracting information features above problem, the problem statement lexical feature S ₂ (Q), characterized in syntactic S ₄ (Q), characterized in lexical semantic S ₃ (Q) and Semantic Syntax wherein S ₅ (Q) four levels The information is analyzed and extracted to lay the foundation for the accuracy of subsequent similarity calculations.

In more embodiments, the feature extraction unit 10 may also analyze and extract the lexical features S ₂ (Q), syntactic features S ₄ (Q), and semantic lexical features S ₃ (Q) of the problem Q according to actual needs. And the semantic syntactic feature S ₅ (Q) can achieve the same technical effect.

Specifically, in the present embodiment, the similarity calculation unit 20 may be configured to calculate the similarity Sim (S _j (Q), S _j (Q _i )) of each type of information feature according to the following formula (1).

Based on the similarity of each type of information feature described above, the similarity Sim(Q,Q _i ) of the problem Qi in the problem and problem library can then be calculated according to the following formula (2):

Where Com(S _j (Q), S _j (Q _i )) is the number of common elements of S _j (Q) and S _j (Q _i ), Num(S _j (Q), S _j (Q _i )) the maximum number of elements S _j (Q) and S _j (Q _i), σ is a smoothing parameter, elements words, combinations of words, the original sense, or a combination of the original sense, j = 2,3,4 or 5.

The smoothing parameter σ is configured according to experience and can be adjusted in the subsequent feedback based on the effect of the smart answer prompt. Setting the smoothing parameter σ can avoid the similarity Sim (S _j ) caused by the difference between Com(S _j (Q), S _j (Q _i )) and Num(S _j (Q), S _j (Q _i )) (Q), S _j (Q _i )) calculation is inaccurate, effectively improving the discrimination between the two to improve the accuracy of the calculation.

In this embodiment, the similarity calculation unit 20 is configured to calculate the similarity Sim (S _j (Q), S _j (Q) of each type of information feature between the problem Q and the problem Q _{i in the} problem library by the above formula (1). _i )), and combines the information of the lexical feature S ₂ (Q), the syntactic feature S ₄ (Q), the semantic lexical feature S ₃ (Q) and the semantic syntactic feature S ₅ (Q), through the above formula (2) Calculate the similarity Sim(Q,Q _i ) of the problem Q _i in the problem and problem library, thereby greatly improving the accuracy and reliability of the similarity calculation of the problem Q and the problem library Q _i .

In some embodiments, the similarity calculation unit 20 can be used to calculate the similarity between the problem Q and the problem Q _i for all the questions Q _i in the problem library, and then perform subsequent operations. Such an embodiment can traverse all of the questions Q _i in the problem library so as not to miss the appropriate answer. In other embodiments, the similarity calculation unit 20 can be used to calculate the similarity of the problem Q to the partial problem for only a part of the problem in the problem library for subsequent processing. In such an embodiment, when the problem library is large, the candidate problem can be selected by preliminary screening (for example, keyword matching, etc.), thereby reducing the amount of calculation and further improving the response speed.

In more embodiments, the similarity calculation unit 20 provided by the present application is not limited to the calculation method by the above formulas (1), (2), and more different calculation methods can be configured to separately calculate the problem Q and the problem library. The similarity of each type of information feature between questions Q _i (S _j (Q), S _j (Q _i )), the similarity of the problem and the problem in the problem library Sim(Q, Q _i ), as long as each pass The similarity of the information features Sim(S _j (Q), S _j (Q _i )) yields the similarity of the problem and the problem in the problem library Sim(Q, Q _i ), which can achieve the same technical effect.

In particular, in the present embodiment, the screening unit 30 answer for (Q, Q _i) of the library have questions Q _i are sorted according to the similarity Sim, sort result forwardmost library problems question Q _i, i.e. The problem closest to question Q, and thus its corresponding answer is provided to the user as a reference for the closest answer.

In this embodiment, the answer screening unit 30 performs sorting and screening based on the similarity Sim(Q, Q _i ), which reduces the time for obtaining the prompt answer, and greatly improves the response speed of the reply.

Preferably, the device for intelligently prompting the answer further includes:

The problem library maintenance unit 40 is configured to classify the problem library into a public question library and a private question library, and to add the question Q and the final answer combination to the private question in response to the final answer determined by the user based on the provided answer. Library. Among them, the public problem database is composed of questions and standard answers in the company's business scope, and the private problem database is customized for each user according to their own needs.

Further preferably, the answer screening unit 30 is configured to sort the problem library problem Q _i according to the similarity order of high priority and private question priority, and select the top result (for example, 3, 5, or 10) questions for the question library Q _i are provided to the user.

In the above embodiment, the answer screening unit 30 further helps to obtain the optimal prompt answer quickly and accurately through reasonable sorting and screening rules, thereby further improving the speed and accuracy of the question answer. The user (for example, customer service) can directly answer the customer's answer by selecting the appropriate answer from the provided prompt answer, or indirectly after answering the prompt answer. Further, through the problem library maintenance unit 40, the public problem library and the private problem library are configured, which not only ensures the accuracy and consistency of the reply, but also realizes the flexibility and personalization of the reply. Private problem libraries can be used only by users themselves or only by groups of users with permissions.

As shown in FIG. 3, the device 300 includes a central processing unit (CPU) 301 that can be loaded from a program stored in a read only memory (ROM) 302 or a program loaded from a storage portion 308 into a random access memory (RAM) 303. Perform various appropriate actions and processes. In the RAM 303, various programs and data required for the operation of the device 300 are also stored. The CPU 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An input/output (I/O) interface 305 is also coupled to bus 304.

The following components are connected to the I/O interface 305: an input portion 306 including a keyboard, a mouse, etc.; an output portion 307 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), and the like, and a storage portion 308 including a hard disk or the like. And a communication portion 309 including a network interface card such as a LAN card, a modem, or the like. The communication section 309 performs communication processing via a network such as the Internet. Driver 310 is also connected to I/O interface 305 as needed. A removable medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory or the like is mounted on the drive 310 as needed so that a computer program read therefrom is installed into the storage portion 308 as needed.

In particular, according to an embodiment of the present disclosure, the process described above with reference to FIG. 1 may be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product comprising a computer program tangibly embodied on a machine readable medium, the computer program comprising program code for performing the method of FIG. In such an embodiment, the computer program can be downloaded and installed from the network via the communication portion 309, and/or installed from the removable medium 311.

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products in accordance with various embodiments of the present invention. In this regard, each block of the flowchart or block diagrams can represent a module, a program segment, or a portion of code that includes one or more logic for implementing the specified. Functional executable instructions. It should also be noted that in some alternative implementations, the functions noted in the blocks may also occur in a different order than that illustrated in the drawings. For example, two successively represented blocks may in fact be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts, can be implemented in a dedicated hardware-based system that performs the specified function or operation. Or it can be implemented by a combination of dedicated hardware and computer instructions.

The units or modules described in the embodiments of the present application may be implemented by software or by hardware. The described unit or module can also be provided in the processor. The names of these units or modules do not in any way constitute a limitation on the unit or module itself.

In another aspect, the present application further provides a computer readable storage medium, which may be a computer readable storage medium included in the apparatus described in the foregoing embodiment, or may exist separately, not A computer readable storage medium that is assembled into the device. A computer readable storage medium stores one or more programs that are used by one or more processors to perform the methods described herein.

The above description is only a preferred embodiment of the present application and a description of the principles of the applied technology. It should be understood by those skilled in the art that the scope of the invention referred to in the present application is not limited to the specific combination of the above technical features, and should also be covered by the above technical features without departing from the inventive concept. Other technical solutions formed by any combination of their equivalent features. For example, the above features are combined with the technical features disclosed in the present application, but are not limited to the technical features having similar functions.

Claims

A method for intelligently prompting an answer, characterized in that the method comprises:

Analyzing and extracting information features of the question Q, the information features including lexical features S 2 (Q), syntactic features S 4 (Q), semantic lexical features S 3 (Q), and semantic syntactic features S 5 (Q);

Similarity information characteristic of each class is calculated between Q and the library have problems question Q i, Q to obtain the similarity Sim problems and problems of libraries question I Q (Q, Q i), wherein said The problem library contains the question Q i and the corresponding answer;

The problem library problem Q i is sorted according to the similarity Sim(Q, Q i ), and the answers of several question database questions Q i with the highest ranking result are provided to the user.
The method for intelligently prompting an answer according to claim 1, wherein the analyzing and extracting information features of the question Q comprises:

Performing word segmentation on the problem Q to obtain a word segmentation sequence S 1 (Q);

Performing part-of-speech tagging on the segmentation sequence S 1 (Q) to obtain a lexical feature S 2 (Q);

Semanticly labeling the lexical feature S 2 (Q) according to a semantic dictionary to obtain a semantic lexical feature S 3 (Q);

Performing adjacent combinations on the lexical feature S 2 (Q) to obtain a syntactic feature S 4 (Q);

The semantic lexical feature S 3 (Q) is adjacently combined to obtain a semantic syntactic feature S 5 (Q).
The method for intelligently prompting an answer according to claim 2, wherein the participle word sequence S 1 (Q) is subjected to part-of-speech tagging, and the lexical feature S 2 (Q) is obtained, including:

Performing part-of-speech tagging on the segmentation sequence S 1 (Q), and retaining nouns, verbs, adjectives and interrogative words therein to obtain a part-of-speech tagging sequence;

The lexical feature S 2 (Q) is the part of speech tagging sequence.
The smart-answer method according to claim 2, wherein said lexical feature according to the semantic lexicon S 2 (Q) semantic annotation, to obtain semantic morphological characteristics S 3 (Q), comprising:

Searching a semantic dictionary, converting words registered in the semantic dictionary in the lexical feature S 2 (Q) into corresponding semantics, and retaining words not registered in the semantic dictionary as meanings, thereby Converting the lexical feature S 2 (Q) to a sense sequence;

The semantic lexical feature S 3 (Q) is the semantic sequence.
The method for intelligently prompting an answer according to any one of claims 1 to 4, characterized in that the similarity Sim (S j (Q), S j (Q i ) of each type of information feature is calculated according to the following formula ):

Where Com(S j (Q), S j (Q i )) is the number of common elements of S j (Q) and S j (Q i ), Num(S j (Q), S j (Q i )) For the maximum number of elements in S j (Q) and S j (Q i ), σ is a smoothing parameter, and the element is a word, a word combination, a sense or a combination of sense, j=2, 3, 4 or 5.
The method for intelligently prompting an answer according to any one of claims 1 to 5, characterized in that the similarity Sim(Q, Q i ) of the problem Q i in the problem and the problem library is calculated according to the following formula:

Where, Sim(S j (Q), S j (Q i )) is the similarity of each type of information feature, j=2, 3, 4 or 5.
The method of intelligently prompting an answer according to any one of claims 1 to 6, wherein the problem library comprises a public problem library and a private problem library.
The method for intelligently answering an answer according to any one of claims 1 to 7, wherein the problem library problem Q i is sorted according to the similarity, and the problem database Q with the highest ranking result is Q. The answer to i is provided to the user, including:

Sort by the order of similarity high priority and private problem first.
The method of intelligently prompting an answer according to any one of claims 1-8, wherein the method further comprises:

Responding to the final answer determined by the user based on the provided answer;

The question Q and the final answer are combined and added to the private question base.
A device for intelligently prompting an answer, characterized in that the device comprises:

A feature extraction unit configured to analyze and extract information features of the question Q, the information features including lexical features S 2 (Q), syntactic features S 4 (Q), semantic lexical features S 3 (Q), and semantic syntactic features S 5 (Q);

Similarity similarity calculation unit, wherein the configuration information for each class is calculated between Q and the problem library have problems of Q I, Q is obtained and the problem of the similarity Sim library have the problem of I Q (Q , Q i ), wherein the problem library contains the question Q i and the corresponding answer;

Answer screening unit, configured according to the similarity Sim (Q, Q i) Q i arsenals problem sorted, the sorted result answer forwardmost Problems arsenals Q i provided to the user.
The apparatus for intelligently answering an answer according to claim 10, wherein the feature extraction unit is configured to analyze and extract information features of the question Q as follows:

Performing word segmentation on the problem Q to obtain a word segmentation sequence S 1 (Q);

The word sequence S 1 (Q) part of speech tagging, to obtain morphological characteristics S 2 (Q);

Semanticly labeling the lexical feature S 2 (Q) according to a semantic dictionary to obtain a semantic lexical feature S 3 (Q);

The morphological characteristics of S 2 (Q) for the adjacent composition obtained syntactic features S 4 (Q); and

The semantic lexical feature S 3 (Q) is adjacently combined to obtain a semantic syntactic feature S 5 (Q).
The apparatus for intelligently answering an answer according to any one of claims 10 to 11, wherein the similarity calculation unit is configured to calculate a similarity Sim (S j ( Q), S j (Q i )):

Wherein, Com (S j (Q) , S j (Q i)) is S j (Q) and S j (Q i) of the common element number, Num (S j (Q) , S j (Q i)) For the maximum number of elements in S j (Q) and S j (Q i ), σ is a smoothing parameter, and the element is a word, a word combination, a sense or a combination of sense, j=2, 3, 4 or 5.
The apparatus for intelligently prompting an answer according to any one of claims 10 to 12, wherein the similarity calculation unit is configured to calculate a similarity Sim of the problem Q i in the problem and the problem library according to the following formula ( Q,Q i ):

Where, Sim(S j (Q), S j (Q i )) is the similarity of each type of information feature, j=2, 3, 4 or 5.
The device for intelligently prompting an answer according to any one of claims 10 to 13, wherein the device further comprises:

a problem library maintenance unit configured to classify the problem library into a public question library and a private question library, and to combine the question Q and the final answer in response to a final answer determined by the user based on the provided answer, adding To the private problem library.
The apparatus for intelligently prompting an answer according to any one of claims 10 to 14, wherein the answer screening unit is configured to sort according to a ranking principle of similarity high priority and private question priority.
A device, characterized in that the device comprises:

One or more processors;

a storage device for storing one or more programs,

The one or more programs are executed by the one or more processors such that the one or more processors implement the method of any of claims 1-9.
A computer readable storage medium having stored thereon a computer program, wherein the program, when executed by a processor, implements the method of any of claims 1-9.