A knowledge question and answer optimal solution selection method and device based on multi-target semantic scoring

Abstract

The application provides a kind of knowledge question and answer optimal solution selection method based on multi-objective semantic score, including obtaining question and splitting to form keyword set, and finding the attribute of each keyword;Determine the standard answer entry in knowledge base and match with keyword set and question respectively, and based on the same standard answer entry, splice two matching results into two-dimensional target score space;According to two-dimensional target score space, construct multi-objective optimal Pareto solution set and form hierarchical structure;According to keyword attribute, calculate dynamic keyword proportion, if greater than preset threshold, select corresponding solution from Pareto layer to merge to Pareto front;Calculate the score of each solution after merging Pareto front, and output the highest score solution as the optimal solution. By implementing the application, static and dynamic keywords in the question can be automatically identified, semantic retrieval path is constructed hierarchically, combined with Pareto optimal mechanism and hybrid scoring strategy, and finally the answer with the optimal comprehensive quality is output.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method and apparatus for selecting the optimal solution for knowledge question answering based on multi-objective semantic scoring. Background Technology

[0002] With the rapid development of knowledge-based question-answering systems, intelligent customer service platforms, and industry knowledge management systems, intelligent retrieval based on semantic similarity has gradually become a key capability in data intelligence applications. Traditional text retrieval methods mostly rely on vector matching mechanisms based on the overall question statement. When faced with complex or semantically redundant question descriptions, they are prone to problems such as insufficient matching, inaccurate understanding of context, and overly simplistic search results. This is especially true when dealing with complex questions in open domains or refined business-oriented consulting tasks, where questions often imply multiple sub-intents and multi-dimensional knowledge requirements.

[0003] However, on the one hand, existing answer selection methods lack the ability to structurally decompose the question and identify key elements, resulting in low coverage of search results, irrelevant entries, and an inability to form a stable and effective answer set. On the other hand, existing answer selection methods typically rely on a single ranking approach, making it difficult to comprehensively consider the trade-offs between multiple matching features (such as maximizing similarity scores, rule-based scoring, or a few heuristic strategies). This "single-point optimal" mechanism has significant limitations in multi-objective optimization scenarios, potentially missing marginal but valuable candidate answers, thus affecting the rationality and coverage of the final results.

[0004] Furthermore, in the semantic composition of a question, different keywords play inconsistent roles in knowledge retrieval. Some keywords are considered static keywords due to their stable business meanings, while others are considered dynamic keywords because they change with the question scenario, user intent, or context (such as conditional descriptions, action verbs, and time states). These dynamic keywords better reflect the current focus of the question. However, existing answer selection methods do not perform structural classification and hierarchical modeling of keywords, resulting in a lack of fine-grained control over semantic matching and limited relevance judgment capabilities.

[0005] Therefore, there is an urgent need for a method that can combine keyword structural attributes for classification processing and introduce a multi-objective optimal solution mechanism for solution set modeling and scoring optimization, so as to improve the response capability, answer accuracy and system intelligence of the answer generation system in complex semantic and multi-objective matching scenarios. Summary of the Invention

[0006] The technical problem to be solved by the embodiments of the present invention is to provide a method and apparatus for selecting the optimal solution of knowledge question answering based on multi-objective semantic scoring, which can automatically identify static and dynamic keywords in the question, construct semantic retrieval paths in layers, combine Pareto optimality mechanism and hybrid scoring strategy, and finally output the answer with the best overall quality.

[0007] To address the aforementioned technical problems, embodiments of the present invention provide a method for selecting the optimal solution in knowledge question answering based on multi-objective semantic scoring, the method comprising the following steps:

[0008] S1. Obtain the user's input question and break it down into multiple keywords to form a keyword set. Further, find the attributes of each keyword from a preset keyword attribute dictionary; wherein, the attributes include static and dynamic.

[0009] S2. Determine a knowledge base containing multiple standard answer entries, and match the keyword set with each standard answer entry in the knowledge base to obtain a keyword set matching result. Also, match the question with each standard answer entry in the knowledge base to obtain a question matching result. Furthermore, concatenate the matching results of the keyword set and the matching results of the question corresponding to the same standard answer entry into a corresponding two-dimensional matching score vector to form a corresponding two-dimensional target scoring space.

[0010] S3. Based on the two-dimensional target scoring space, construct a multi-objective optimal Pareto solution set; wherein, the Pareto solution set is a hierarchical structure formed by dominating and sorting the score vectors of all candidate solutions, including the Pareto front and multiple Pareto layers outside it.

[0011] S4. Based on the attributes of each keyword, distinguish dynamic keywords from the multiple keywords to calculate the proportion of dynamic keywords. When it is determined that the proportion is greater than a preset threshold, select one or more Pareto layers from the Pareto solution set. After calculating the acceptability score of each solution in the selected Pareto layer, select the solutions in the selected Pareto layers whose acceptability scores meet the predetermined conditions, and further merge the selected solutions into the Pareto front.

[0012] S5. Statistically calculate the minimum and maximum values ​​in the merged Pareto front as the ideal solution and the negative ideal solution, and calculate the score of each solution in the merged Pareto front other than the ideal solution and the negative ideal solution, and further output the solution with the highest score as the optimal solution.

[0013] In step S1, the keyword set is obtained by using natural language processing tools to analyze the question. The expression obtained after word segmentation is: in, Indicates from the question The number of keywords extracted; Indicates the first One keyword.

[0014] In step S1, the formula is used. Find the attributes of each keyword; among them,

[0015] Keywords Attributes; Indicates static; Indicates dynamic; the expression for the keyword attribute dictionary is: ; and These represent keyword attribute dictionaries. The Middle Keywords and their attributes.

[0016] Specifically, step S2 includes:

[0017] Determined to contain The knowledge base of standard answer entries ;in, ; and They represent the first The first standard answer entry and the first One standard answer entry; This represents the total number of standard answer entries, and it is a positive integer greater than 1.

[0018] Using the cosine similarity function The keyword set With the knowledge base In Each standard answer entry was matched to obtain the keyword set matching results. and the issues mentioned With the knowledge base In Each standard answer entry was matched to obtain the question matching results. ;in, ; ; ;

[0019] Matching results of the keyword set and the matching results of the aforementioned issues The matching results for the same standard answer item are concatenated to form a corresponding two-dimensional matching score vector. All the splicing results are then combined to form a two-dimensional target scoring space. ;in, , The score vectors under the two matching paths; .

[0020] Specifically, step S3 includes:

[0021] Based on the two-dimensional target scoring space Construct a multi-objective optimal Pareto solution set; where, for any two and Their score vectors are respectively , If the formula is satisfied Then determine Dominate of ;

[0022] The score vectors of all candidate solutions are non-dominated and sorted to form multiple levels in the Pareto solution set; wherein, all levels in the Pareto solution set are defined as follows: ; This is the Pareto front, which represents all solutions that are not dominated by any other solution; This is the second Pareto layer, which represents the non-dominated solution after removing the first layer; For the first Pareto layers, representing the removal of the first through the second Pareto layers. Non-dominated solutions after the layer; This represents the total number of floors.

[0023] Specifically, step S4 includes:

[0024] Based on the attributes of each keyword, dynamic keywords are distinguished from the multiple keywords to form a dynamic keyword set. ;in,

[0025] Through formula Calculate the proportion of dynamic keywords ;in, Indicates quantity statistics;

[0026] If the proportion of dynamic keywords Greater than the preset threshold Then, two Pareto layers are selected from the Pareto solution set, including the second Pareto layer. and the third Pareto layer

[0027] Through formula Calculate the second Pareto layer and the third Pareto layer The relaxation dominance score of all solutions ;in, and They represent score vector middle Two dimensions in the second Pareto level The third Pareto layer The value; and They represent score vector middle The maximum values ​​of the two dimensions at the Pareto front;

[0028] Through formula Calculate the second Pareto layer and the third Pareto layer The dynamic keyword semantic fit score of all solutions ;in, Dynamic keyword collection The average semantic vector is used to indicate the semantic similarity between the solution and the dynamic key vector;

[0029] Through formula Calculate the second Pareto layer and the third Pareto layer Acceptability score of all solutions ;in, and Both represent adjustment parameters, and both are preset constants;

[0030] Following the order from high to low, the second Pareto level and the third Pareto layer Acceptability score of all solutions Sort, and start from the second Pareto level. and the third Pareto layer Select the acceptable score from all solutions. Arranged at the front A solution;

[0031] Acceptability score Arranged at the front All solutions are merged into the Pareto front. In order to form the Pareto front after the merger ;in, , Indicates the acceptability score Arranged at the front The set of solutions.

[0032] Specifically, step S5 includes:

[0033] The Pareto Frontier after the merger In the middle, obtain the score vector of all solutions. The solution that maximizes both dimensions is taken as the ideal solution, and the score vector of all solutions is also considered. The solution that minimizes both dimensions is considered the negative ideal solution;

[0034] Through formula The combined Pareto front was calculated. The score for each solution other than the ideal solution and the negative ideal solution; wherein, This represents the combined Pareto front. The third solution other than the ideal solution and the negative ideal solution The score for each solution; Indicates the first The distance between each solution and the negative ideal solution; Indicates the first The distance between each solution and the ideal solution;

[0035] All scores obtained Sort the solutions from highest to lowest score and output the solution with the highest score as the optimal solution.

[0036] This invention also provides a knowledge question answering optimal solution selection device based on multi-objective semantic scoring, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the aforementioned knowledge question answering optimal solution selection method based on multi-objective semantic scoring.

[0037] Implementing the embodiments of the present invention has the following beneficial effects:

[0038] This invention decomposes the problem into two categories of keywords: static and dynamic, and performs semantic matching with an unstructured text knowledge base. Combining Pareto front modeling and a conditional preservation strategy for non-dominated solutions, and integrating semantic fit and dominance relaxation as dual indicators, it can effectively construct a candidate answer set with strong coverage and semantic rationality. Finally, it uses the TOPSIS scoring method to output the optimal solution with stable structure and strong semantic relevance, significantly improving the accuracy of intelligent response and the adaptability of the system in complex problem scenarios. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, obtaining other drawings based on these drawings without creative effort still falls within the scope of the present invention.

[0040] Figure 1 This is a flowchart of a knowledge question answering optimal solution selection method based on multi-objective semantic scoring, provided for an embodiment of the present invention. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings.

[0042] like Figure 1 As shown in the figure, an optimal solution selection method for knowledge question answering based on multi-objective semantic scoring is provided in an embodiment of the present invention. The method includes the following steps:

[0043] Step S1: Obtain the user's input question and break it down into multiple keywords to form a keyword set. Then, find the attributes of each keyword from a preset keyword attribute dictionary. The attributes include static and dynamic attributes.

[0044] The specific process is as follows: First, based on the question input by the user... Using natural language processing tools to address the problem Word segmentation is performed to obtain a keyword set; where the expression for this keyword set is: Indicates from the question The number of keywords extracted; Indicates the first One keyword.

[0045] Secondly, construct a keyword attribute dictionary. ;in, and These represent keyword attribute dictionaries. The Middle Keywords and their attributes.

[0046] Finally, regarding keywords Through this keyword attribute dictionary Determine its attributes.

[0047] For example, through formula Find the attributes of each keyword; among them, Keywords Attributes; Indicates static; Indicates dynamic.

[0048] Therefore, all keywords are divided into two subsets:

[0049] Static keyword set:

[0050] Dynamic keyword collection: .

[0051] Step S2: Determine a knowledge base containing multiple standard answer entries, and match the keyword set with each standard answer entry in the knowledge base to obtain a keyword set matching result. Also, match the question with each standard answer entry in the knowledge base to obtain a question matching result. Furthermore, concatenate the matching results of the keyword set and the matching results of the question with the same standard answer entry to form a corresponding two-dimensional matching score vector to form a corresponding two-dimensional target scoring space.

[0052] The specific process is as follows: First, determine the contents The knowledge base of standard answer entries ;in, ; and They represent the first The first standard answer entry and the first One standard answer entry; This represents the total number of standard answer entries, and it is a positive integer greater than 1.

[0053] Secondly, the overall problem and keyword set The knowledge base is jointly input, and the cosine similarity function is used. , keyword set With knowledge base In Each standard answer entry was matched to obtain the keyword set matching results. and the problem With knowledge base In Each standard answer entry was matched to obtain the question matching results. ;in, ; ;

[0054] Next, in the keyword set matching results and problem matching results The matching results for the same standard answer item are concatenated to form a corresponding two-dimensional matching score vector. All the splicing results are then combined to form a two-dimensional target scoring space. ;in, , The score vectors under the two matching paths are used for subsequent Pareto ranking; .

[0055] It should be noted that the two matching paths represent keyword sets respectively. With knowledge base In The path and question for matching each standard answer entry. With knowledge base In The path for matching the standard answer entries.

[0056] Step S3: Based on the two-dimensional target scoring space, construct the multi-objective optimal Pareto solution set; wherein, the Pareto solution set is a multi-level structure formed by dominating and sorting the score vectors of all candidate solutions, including the Pareto front and multiple Pareto layers outside it.

[0057] The specific process is as follows: First, based on the two-dimensional target scoring space... Construct a multi-objective optimal Pareto solution set; where, for any two and Their score vectors are respectively , If the formula is satisfied Then determine Dominate of

[0058] Then, the score vectors of all candidate solutions are non-dominated to form multiple levels in the Pareto solution set; where each level in the Pareto solution set is defined as... ; This is the Pareto front, which represents all solutions that are not dominated by any other solution; This is the second Pareto layer, which represents the non-dominated solution after removing the first layer; For the first Pareto layers, representing the removal of the first through the second Pareto layers. Non-dominated solutions after the layer; This represents the total number of floors.

[0059] Step S4: Based on the attributes of each keyword, distinguish dynamic keywords from the multiple keywords to calculate the proportion of dynamic keywords. When the proportion is determined to be greater than a preset threshold, select one or more Pareto layers from the Pareto solution set. After calculating the acceptability score of each solution in the selected Pareto layer, select the solutions whose acceptability scores meet the predetermined conditions in the selected Pareto layers, and further merge the selected solutions into the Pareto front.

[0060] The specific process is as follows: First, based on the attributes of each keyword in step S1, a dynamic keyword set is obtained from multiple keywords.

[0061] Secondly, through the formula Calculate the proportion of dynamic keywords ;in, This represents a quantity counter.

[0062] Then, if the proportion of dynamic keywords Greater than the preset threshold Then, two Pareto layers are selected from the Pareto solution set, including the second Pareto layer. and the third Pareto layer

[0063] Through formula Calculate the second Pareto layer and the third Pareto layer The relaxation dominance score of all solutions ;in, and They represent score vector middle Two dimensions in the second Pareto level The third Pareto layer The value; and They represent score vector middle The maximum values ​​of the two dimensions at the Pareto front;

[0064] Through formula Calculate the second Pareto layer and the third Pareto layer The dynamic keyword semantic fit score of all solutions ;in, Dynamic keyword collection The average semantic vector is used to indicate the semantic similarity between the solution and the dynamic key vector;

[0065] Through formula Calculate the second Pareto layer and the third Pareto layer Acceptability score of all solutions ;in, and Both represent adjustment parameters, and both are preset constants;

[0066] Following the order from high to low, the second Pareto level and the third Pareto layer Acceptability score of all solutions Sort, and start from the second Pareto level. and the third Pareto layer Select the acceptable score from all solutions. Arranged at the front A solution; that is, for All solutions are calculated Values, sorted by score, and taking the top. One as a supplementary solution set .

[0067] This shows the proportion of dynamic keywords. Preset threshold At that time, we attempted to introduce supplementary answers from non-Pareto front solutions.

[0068] Finally, the acceptability score will be calculated. Arranged at the front All solutions are merged into the Pareto front. In order to form the Pareto front after the merger ;in, , Indicates the acceptability score Arranged at the front The set of solutions.

[0069] Step S5: Statistically calculate the minimum and maximum values ​​in the merged Pareto front as the ideal solution and the negative ideal solution, and calculate the score of each solution in the merged Pareto front other than the ideal solution and the negative ideal solution, and further output the solution with the highest score as the optimal solution.

[0070] The specific process is as follows: First, at the merged Pareto front... In the middle, obtain the score vector of all solutions. The solution that maximizes both dimensions is taken as the ideal solution, and the score vector of all solutions is also considered. The solution that minimizes both dimensions is considered the negative ideal solution.

[0071] Secondly, through the formula Calculate the Pareto front after merging The score for each solution other than the ideal solution and the negative ideal solution; wherein, Indicates the Pareto front after the merger The third solution, excluding the ideal solution and the negative ideal solution The score for each solution; Indicates the first The distance between the solution and the negative ideal solution; Indicates the first The distance between each solution and the ideal solution.

[0072] Finally, all the scores obtained Sort the solutions from highest to lowest score and output the solution with the highest score as the optimal solution.

[0073] Corresponding to the knowledge question answering optimal solution selection method based on multi-objective semantic scoring provided in the embodiments of the present invention, the embodiments of the present invention also provide a knowledge question answering optimal solution selection device based on multi-objective semantic scoring, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the knowledge question answering optimal solution selection method based on multi-objective semantic scoring provided in the embodiments of the present invention. For details, please refer to the relevant content above, which will not be repeated here.

[0074] Implementing the embodiments of the present invention has the following beneficial effects:

[0075] This invention decomposes the problem into two categories of keywords: static and dynamic, and performs semantic matching with an unstructured text knowledge base. Combining Pareto front modeling and a conditional preservation strategy for non-dominated solutions, and integrating semantic fit and dominance relaxation as dual indicators, it can effectively construct a candidate answer set with strong coverage and semantic rationality. Finally, it uses the TOPSIS scoring method to output the optimal solution with stable structure and strong semantic relevance, significantly improving the accuracy of intelligent response and the adaptability of the system in complex problem scenarios.

[0076] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as ROM / RAM, disk, optical disk, etc.

[0077] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for selecting the optimal solution in knowledge question answering based on multi-objective semantic scoring, characterized in that, The method includes the following steps: S1. Obtain the user's input question and break it down into multiple keywords to form a keyword set. Further, find the attributes of each keyword from a preset keyword attribute dictionary; wherein, the attributes include static and dynamic. S2. Determine a knowledge base containing multiple standard answer entries, and match the keyword set with each standard answer entry in the knowledge base to obtain a keyword set matching result. Also, match the question with each standard answer entry in the knowledge base to obtain a question matching result. Furthermore, concatenate the matching results of the keyword set and the matching results of the question corresponding to the same standard answer entry into a corresponding two-dimensional matching score vector to form a corresponding two-dimensional target scoring space. S3. Based on the two-dimensional target scoring space, construct a multi-objective optimal Pareto solution set; wherein, the Pareto solution set is a hierarchical structure formed by dominating and sorting the score vectors of all candidate solutions, including the Pareto front and multiple Pareto layers outside it. S4. Based on the attributes of each keyword, distinguish dynamic keywords from the multiple keywords to calculate the proportion of dynamic keywords. When it is determined that the proportion is greater than a preset threshold, select one or more Pareto layers from the Pareto solution set. After calculating the acceptability score of each solution in the selected Pareto layer, select the solutions in the selected Pareto layers whose acceptability scores meet the predetermined conditions, and further merge the selected solutions into the Pareto front. S5. Statistically calculate the minimum and maximum values ​​in the merged Pareto front as the ideal solution and the negative ideal solution, and calculate the score of each solution in the merged Pareto front other than the ideal solution and the negative ideal solution, and further output the solution with the highest score as the optimal solution.

2. The optimal solution selection method for knowledge question answering based on multi-objective semantic scoring as described in claim 1, characterized in that, In step S1, the keyword set is obtained by using natural language processing tools to analyze the question. The expression obtained after word segmentation is: in, Indicates from the question The number of keywords extracted; Indicates the first One keyword.

3. The optimal solution selection method for knowledge question answering based on multi-objective semantic scoring as described in claim 2, characterized in that, In step S1, through the formula Find the attributes of each keyword; among them, Keywords Attributes; Indicates static; Indicates dynamic; the expression for the keyword attribute dictionary is: ; and These represent keyword attribute dictionaries. The Middle Keywords and their attributes.

4. The optimal solution selection method for knowledge question answering based on multi-objective semantic scoring as described in claim 3, characterized in that, Step S2 specifically includes: Determined to contain The knowledge base of standard answer entries ;in, ; and They represent the first The first standard answer entry and the first One standard answer entry; This represents the total number of standard answer entries, and it is a positive integer greater than 1. Using the cosine similarity function The keyword set With the knowledge base In Each standard answer entry was matched to obtain the keyword set matching results. and the issues mentioned With the knowledge base In Each standard answer entry was matched to obtain the question matching results. ;in, ; ; ; Matching results of the keyword set and the matching results of the aforementioned issues The matching results for the same standard answer item are concatenated to form a corresponding two-dimensional matching score vector. All the splicing results are then combined to form a two-dimensional target scoring space. ;in, , The score vectors under the two matching paths; .

5. The optimal solution selection method for knowledge question answering based on multi-objective semantic scoring as described in claim 4, characterized in that, Step S3 specifically includes: Based on the two-dimensional target scoring space Construct a multi-objective optimal Pareto solution set; where, for any two and Their score vectors are respectively , If the formula is satisfied Then determine Dominate of ; The score vectors of all candidate solutions are non-dominated and sorted to form multiple levels in the Pareto solution set; wherein, all levels in the Pareto solution set are defined as follows: ; This is the Pareto front, which represents all solutions that are not dominated by any other solution; This is the second Pareto layer, which represents the non-dominated solution after removing the first layer; For the first Pareto layers, representing the removal of the first through the second Pareto layers. Non-dominated solutions after the layer; This represents the total number of floors.

6. The optimal solution selection method for knowledge question answering based on multi-objective semantic scoring as described in claim 5, characterized in that, Step S4 specifically includes: Based on the attributes of each keyword, dynamic keywords are distinguished from the multiple keywords to form a dynamic keyword set. ;in, Through formula Calculate the proportion of dynamic keywords ;in, Indicates quantity statistics; If the proportion of dynamic keywords Greater than the preset threshold Then, two Pareto layers are selected from the Pareto solution set, including the second Pareto layer. and the third Pareto layer Through formula Calculate the second Pareto layer and the third Pareto layer The relaxation dominance score of all solutions ;in, and They represent score vector middle Two dimensions in the second Pareto level The third Pareto layer The value; and They represent score vector middle The maximum values ​​of the two dimensions at the Pareto front; Through formula Calculate the second Pareto layer and the third Pareto layer The dynamic keyword semantic fit score of all solutions ;in, Dynamic keyword collection The average semantic vector is used to indicate the semantic similarity between the solution and the dynamic key vector; Through formula Calculate the second Pareto layer and the third Pareto layer Acceptability score of all solutions ;in, and Both represent adjustment parameters, and both are preset constants; Following the order from high to low, the second Pareto level and the third Pareto layer Acceptability score of all solutions Sort, and start from the second Pareto level. and the third Pareto layer Select the acceptable score from all solutions. Arranged at the front A solution; Acceptability score Arranged at the front All solutions are merged into the Pareto front. In order to form the Pareto front after the merger ;in, , Indicates the acceptability score Arranged at the front The set of solutions.

7. The optimal solution selection method for knowledge question answering based on multi-objective semantic scoring as described in claim 6, characterized in that, Step S5 specifically includes: The Pareto Frontier after the merger In the middle, obtain the score vector of all solutions. The solution that maximizes both dimensions is taken as the ideal solution, and the score vector of all solutions is also considered. The solution that minimizes both dimensions is considered the negative ideal solution. Through formula The combined Pareto front was calculated. The score for each solution other than the ideal solution and the negative ideal solution; wherein, This represents the combined Pareto front. The third solution other than the ideal solution and the negative ideal solution The score for each solution; Indicates the first The distance between each solution and the negative ideal solution; Indicates the first The distance between each solution and the ideal solution; All scores obtained Sort the solutions from highest to lowest score and output the solution with the highest score as the optimal solution.

8. A knowledge question answering optimal solution selection device based on multi-objective semantic scoring, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the knowledge question answering optimal solution selection method based on multi-objective semantic scoring as described in any one of claims 1-7.

Images