Confusion knowledge point detection method and device, electronic equipment and computer storage medium

Abstract

The embodiment of the application provides a method and device for detecting confusing knowledge points, electronic equipment and computer storage medium, wherein the method for detecting confusing knowledge points comprises the following steps: determining a candidate set of confusing knowledge points according to the keyword coincidence degree between a plurality of knowledge points in a knowledge base; calculating the semantic similarity between a first knowledge point and a second knowledge point in the candidate set of confusing knowledge points; and if the semantic similarity is higher than a preset threshold similarity, determining that the first knowledge point and the second knowledge point are confusing knowledge points. The keyword coincidence degree can be used to quickly screen out the candidate set of confusing knowledge points, and then the semantic similarity can be used to more accurately identify the confusing knowledge points, so that the identification accuracy and speed are taken into account.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of artificial intelligence, and in particular to a method and device for detecting confusing knowledge points, an electronic device, and a computer storage medium. BACKGROUND

[0002] In many fields, for example, when purchasing goods on the Internet, conducting business through a webpage, etc., users have various questions that need to be consulted. In order to improve the answering efficiency and reduce the labor cost, many common questions can be intelligently pushed answers after the user inputs a question that wants to be consulted. In the implementation process of the above scheme, a knowledge base usually needs to be established, which contains multiple questions and corresponding multiple answers. However, some questions are very similar, and the device may identify incorrectly in the identification process, thereby pushing an incorrect answer to the user. These similar questions are confusing questions. In order to reduce the influence of the confusing questions, the confusing questions need to be selected out and processed in the knowledge base, so that the device can correctly identify these questions. In the related art, the identification accuracy is low and the identification speed is slow in the process of identifying the confusing questions. SUMMARY

[0003] Therefore, embodiments of the present application provide a method and device for detecting confusing knowledge points, an electronic device, and a computer storage medium to solve some or all of the above problems.

[0004] According to a first aspect of embodiments of the present application, a method for detecting confusing knowledge points is provided, including: determining a candidate set of confusing knowledge points according to the keyword coincidence degree between multiple knowledge points in a knowledge base; calculating the semantic similarity between a first knowledge point and a second knowledge point in the candidate set of confusing knowledge points; and if the semantic similarity is higher than a preset threshold similarity, determining that the first knowledge point and the second knowledge point are confusing knowledge points.

[0005] According to a second aspect of embodiments of the present application, a device for detecting confusing knowledge points is provided, including: a keyword module configured to determine at least one candidate knowledge point pair according to the keyword coincidence degree of at least one knowledge point pair in a knowledge base; a semantic module configured to determine the semantic similarity of each candidate knowledge point pair in the at least one candidate knowledge point pair; and a confusion module configured to determine at least one confusing knowledge point pair from the at least one candidate knowledge point pair according to the semantic similarity.

[0006] According to a third aspect of embodiments of the present application, an electronic device is provided, including a processor, a memory, a communication interface, and a communication bus, the processor, the memory, and the communication interface complete communication with each other through the communication bus; the memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the method for detecting confusing knowledge points according to the first aspect.

[0007] According to a fourth aspect of the embodiments of the present application, a computer storage medium is provided, and the computer storage medium stores a determination program, and the program is executed by a processor to implement the detection method of the confusing knowledge points according to the first aspect.

[0008] The detection method of the confusing knowledge points, the device, the electronic equipment and the computer storage medium provided by the embodiments of the present application can determine the candidate set of the confusing knowledge points according to the keyword coincidence degree between the multiple knowledge points in the knowledge base, calculate the semantic similarity between the first knowledge point and the second knowledge point in the candidate set of the confusing knowledge points, and determine the first knowledge point and the second knowledge point as the confusing knowledge points if the semantic similarity is higher than the preset threshold similarity. The keyword coincidence degree can be used to quickly screen out the candidate set of the confusing knowledge points, and then the semantic similarity can be used to more accurately identify the confusing knowledge points, so that the identification accuracy and the identification speed are taken into account. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments described in the embodiments of the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0010] Figure 1 A scene schematic diagram of a detection method of confusing knowledge points provided by the first embodiment of the present application;

[0011] Figure 2 A flowchart of a detection method of confusing knowledge points provided by the first embodiment of the present application;

[0012] Figure 3 A schematic diagram of a knowledge base provided by the first embodiment of the present application;

[0013] Figure 4 An application scene schematic diagram of a knowledge base provided by the first embodiment of the present application;

[0014] Figure 5 A flowchart of a detection method of confusing knowledge points provided by the second embodiment of the present application;

[0015] Figure 6 An application scene schematic diagram of a detection method of confusing knowledge points provided by the second embodiment of the present application;

[0016] Figure 7 A structural block diagram of a detection device of confusing knowledge points provided by the third embodiment of the present application;

[0017] Figure 8 A structural schematic diagram of an electronic equipment provided by the fourth embodiment of the present application. Detailed Implementation

[0018] To enable those skilled in the art to better understand the technical solutions in the embodiments of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art should fall within the protection scope of the embodiments of this application.

[0019] The specific implementation of the embodiments of this application will be further described below with reference to the accompanying drawings.

[0020] Example 1

[0021] This application provides a method for detecting obfuscated knowledge points, applied to electronic devices. For ease of understanding, the application scenario of the method for detecting obfuscated knowledge points provided in this application is described below, referring to… Figure 1 As shown, Figure 1 This is a schematic diagram of a scenario for a method for detecting confusing knowledge points provided in Embodiment 1 of this application. Figure 1 The scenario shown includes an electronic device 101; the electronic device 101 may be a device that performs the method for detecting obfuscated knowledge points provided in Embodiment 1 of this application.

[0022] Electronic device 101 can be a terminal device such as a smart speaker, smartphone, tablet computer, or laptop computer. Electronic device 101 can also be a cloud device, such as a server, relay device, or end-to-end (D2D) device.

[0023] like Figure 1As shown, the electronic device 101 comprises a knowledge base, which can include at least one knowledge point, and one knowledge point can include one document. For example, the knowledge base can be a question and answer base, which includes at least two questions and answers corresponding to the questions, one or more related questions can form a knowledge point (i.e. one document), one knowledge point can include one answer and one or more questions corresponding to the answer. Two knowledge points form a knowledge point pair, and the keyword coincidence degree of at least one knowledge point pair (multiple knowledge points) can be calculated by calculating the keyword coincidence degree of two knowledge points. According to the keyword coincidence degree of at least one knowledge point pair, the electronic device 101 can determine at least one candidate knowledge point pair and obtain a candidate confusion knowledge point set, and at least one confusion knowledge point pair can be determined by using the semantic similarity of each candidate knowledge point pair in at least one candidate knowledge point pair. By using the keyword coincidence degree and the semantic similarity, the knowledge base is double-screened, and the confusion knowledge point can be determined more quickly and accurately.

[0024] In combination Figure 1 The scene shown in the scene, the embodiment of the present application provides a confusion knowledge point detection method, it needs to be explained that, Figure 1 Only one application scenario of the confusion knowledge point detection method provided by the embodiment of the present application, and it does not mean that the confusion knowledge point detection method must be applied to Figure 1 The scene shown in the scene, refer to Figure 2 As shown, Figure 2 A flow chart of a confusion knowledge point detection method provided by the embodiment of the present application, the method comprises the following steps:

[0025] Step 201, according to the keyword coincidence degree between multiple knowledge points in the knowledge base, determine the confusion knowledge point candidate set.

[0026] It needs to be explained that the keyword coincidence degree between multiple knowledge points can include the keyword coincidence degree of at least one knowledge point pair in the knowledge base, that is, the keyword coincidence degree of two knowledge points in multiple knowledge points. The confusion knowledge point candidate set can include at least one candidate knowledge point pair determined.

[0027] The knowledge base can include multiple knowledge points, that is, at least two knowledge points, and two knowledge points in multiple knowledge points can form a knowledge point pair, for example, the knowledge base includes 10 knowledge points, and each two knowledge points can form a knowledge point pair, which can form C 10 2 = 45 knowledge point pairs. For example, as shown in Figure 3 Figure 3 ​A schematic diagram of a knowledge base provided for Embodiment One of the present application. The knowledge base can be a question and answer base. A knowledge point can be a document. A knowledge point can include at least one question and a corresponding answer. In a specific implementation, a knowledge point can include a standard question and an extended question (also referred to as a similar question). The standard question can be one, and the extended question can be one or more. The standard question and the extended question can have the same meaning, but different expressions. The standard question and the multiple extended questions correspond to the same answer. Of course, a knowledge point can also include multiple answers. This is only an exemplary illustration.

[0028] A specific example is given to illustrate how to determine a candidate knowledge point pair. Optionally, the method for determining a candidate set of confusion knowledge points based on the keyword overlap degree between multiple knowledge points in the knowledge base includes: determining the keywords corresponding to each knowledge point; using the keywords to perform matching to obtain at least two knowledge points with a keyword overlap degree higher than a preset overlap degree to form a candidate set of confusion knowledge points. It should be noted that the keyword overlap degree can indicate how many same keywords are included in the two knowledge points in a knowledge point pair. Optionally, the Term Frequency-Inverse Document Frequency (TF-IDF) method can be used to calculate the keywords corresponding to each knowledge point. For two knowledge points (i.e., a knowledge point pair), the number of same keywords can be counted. The keyword overlap degree is determined based on the number of same keywords. The keyword overlap degree is positively correlated with the number or proportion of same keywords of the two knowledge points. For example, the number of same keywords can be determined as the keyword overlap degree. Alternatively, the proportion of the number of same keywords in the total number of keywords of the two knowledge points can be determined as the keyword overlap degree. As long as the value calculated based on the number of same keywords is positively correlated with the number or proportion of same keywords, it can be used as the keyword overlap degree. Using the keyword overlap degree, the candidate knowledge point pair can be quickly determined, improving the efficiency. Exemplarily, the keyword overlap degree can be determined as follows: based on the correspondence between the keywords and the knowledge points, at least two keywords of the knowledge points are used to perform retrieval; and the keyword overlap degree of at least one knowledge point pair is determined based on the retrieval result. It should be noted that the keywords and the knowledge points can form an index indicating the correspondence between the keywords and the knowledge points. For example, for keyword A, using the index, all knowledge points including keyword A can be directly determined. Using the correspondence between the keywords and the knowledge points to perform retrieval further improves the efficiency of calculating the keyword overlap degree.

[0029] It should be noted that the candidate set of confusing knowledge points can include at least one candidate knowledge point pair, each candidate knowledge point pair including two knowledge points with keyword coincidence degrees higher than a preset coincidence degree. The keyword coincidence degree of the at least one knowledge point pair can be determined according to keywords of at least two knowledge points of the knowledge base, and the at least two knowledge points can form the at least one knowledge point pair; according to the keyword coincidence degree of the at least one knowledge point pair, the knowledge point pair with a keyword coincidence degree higher than the preset coincidence degree is determined as the at least one candidate knowledge point pair in the at least one knowledge point pair. Alternatively, the at least one knowledge point pair can be sorted in descending order of keyword coincidence degrees, and the knowledge point pair with a keyword coincidence degree greater than or equal to the preset coincidence degree is determined as the candidate knowledge point pair, and the candidate set of confusing knowledge points is obtained. It should be noted that the preset coincidence degree can be set according to specific conditions, the at least one knowledge point pair is screened by using the preset coincidence degree to obtain the candidate knowledge point pair, the screening condition can be more flexibly controlled, and the at least one knowledge point pair is sorted in descending order of keyword coincidence degrees, which can improve the efficiency of screening. Alternatively, in a specific implementation manner, the at least one knowledge point pair can be sorted according to keyword coincidence degrees, and the top X knowledge point pairs are determined as the candidate knowledge point pairs, X being an integer greater than 0. Of course, this is only an example.

[0030] Step 202, calculating a semantic similarity between the first knowledge point and the second knowledge point in the candidate set of confusing knowledge points.

[0031] The first knowledge point and the second knowledge point are a knowledge point pair, and the semantic similarity can be calculated for each knowledge point pair in the candidate set of confusing knowledge points, so as to determine the semantic similarity of each candidate knowledge point pair in the candidate set of confusing knowledge points.

[0032] The semantic similarity can indicate a degree of semantic similarity between the two knowledge points in the knowledge point pair, and the greater the semantic similarity, the more similar the meanings of the two knowledge points. Of course, in combination with the knowledge base shown in Figure 3 The semantic similarity can also represent a degree of semantic similarity between two problems.

[0033] Optionally, the knowledge points in the knowledge base include a plurality of questions and answers corresponding to the questions; accordingly, in the set of candidate confused knowledge points, the semantic similarity between the first knowledge point and the second knowledge point is calculated, including: obtaining a plurality of questions corresponding to the first knowledge point and a plurality of questions corresponding to the second knowledge point; calculating the pairwise semantic similarity between the plurality of questions corresponding to the first knowledge point and the plurality of questions corresponding to the second knowledge point; and determining the semantic similarity between the first knowledge point and the second knowledge point according to the pairwise semantic similarity. Optionally, the similarity of two questions in different knowledge points in each question pair in the candidate knowledge point pair can be calculated to obtain at least one question pair similarity; and the semantic similarity of the candidate knowledge point pair is determined according to the at least one question pair similarity. It should be noted that the two questions in each question pair are in different knowledge points, and the similarity of the two questions can be calculated by using a semantic recognition model, combined with Figure 3 The knowledge point is a set of at least one question, and the similarity between the two questions in each question pair in the two knowledge points can more accurately calculate the semantic similarity of the candidate knowledge point pair.

[0034] Here, two examples are given to illustrate how to calculate the semantic similarity of the candidate knowledge point pair. Optionally, in the first example, the semantic similarity of the candidate knowledge point pair is determined according to the at least one question pair similarity, including: calculating the average value of the similarity of the first N question pairs in descending order of similarity, and determining the average value of the similarity as the semantic similarity of the candidate knowledge point pair, N being an integer greater than 0. Only the average value of the similarity of the first N question pairs is taken, and the question pairs with smaller similarity are not considered, so that the candidate knowledge point pair can be more accurately determined whether it contains question pairs with semantic similarity. Optionally, in the second example, the semantic similarity of the candidate knowledge point pair is determined according to the at least one question pair similarity, including: taking the average value of the similarity of at least one question pair contained in the knowledge point pair as the semantic similarity of the knowledge point pair. The average value of the similarity of at least one question pair can more comprehensively reflect the semantic similarity between the candidate knowledge point pairs.

[0035] Step 203: If the semantic similarity is higher than the preset similarity, determining that the first knowledge point and the second knowledge point are confused knowledge points.

[0036] The first knowledge point and the second knowledge point are easy to confuse, and can be referred to as a confused knowledge point pair. Each knowledge point pair in the confused knowledge point candidate set can be calculated to determine the semantic similarity, and the candidate knowledge point pair with a semantic similarity higher than a preset similarity is determined as a confused knowledge point pair. It should be noted that the confused knowledge point pair includes two knowledge points that are easy to confuse, and the preset similarity can be set according to specific conditions, which is not limited in the present application. The confused knowledge point pair is determined in at least one candidate knowledge point pair by using the semantic similarity, which can more accurately determine the knowledge point pair with semantic confusion, and improves the accuracy of determining the confused knowledge point pair. Optionally, in an embodiment, the semantic similarity between the first knowledge point and the second knowledge point is determined according to the semantic similarity between each two, which includes: calculating the average value of the similarity of the first N question pairs in the order of similarity from large to small, and determining the average value of the similarity as the semantic similarity between the first knowledge point and the second knowledge point, N is an integer greater than 0. Of course, this is only an example to illustrate how to determine the semantic similarity.

[0037] Based on the method described in steps 201-203, in order to further improve the accuracy of determining the confused knowledge point pair, the confused knowledge point pair can be further screened after steps 201-203. Here, two specific examples are listed for further explanation.

[0038] Optionally, in the first example, after determining that the first knowledge point and the second knowledge point are the confusion knowledge points, the method further comprises: determining keyword coincidence degrees of the first M question pairs in the question pairs composed of the multiple questions corresponding to the first knowledge point and the multiple questions corresponding to the second knowledge point in the order of the similarity degrees from large to small, and determining the expansion coincidence degree between the first knowledge point and the second knowledge point according to the keyword coincidence degrees of the first M question pairs, M being an integer greater than 0; and if the expansion coincidence degree is greater than a preset expansion coincidence degree, determining that the confusion knowledge point composed of the first knowledge point and the second knowledge point is a high-quality confusion knowledge point. Screening the high-quality confusion knowledge points from the confusion knowledge points, or filtering some knowledge points from the confusion knowledge points, for example, determining the keyword coincidence degrees of the first M question pairs in the at least one question pair in the order of the similarity degrees from large to small, and determining the expansion coincidence degree of the confusion knowledge point pair according to the keyword coincidence degrees of the first M question pairs, M being an integer greater than 0; and filtering the confusion knowledge point pairs with the expansion coincidence degrees less than or equal to the preset expansion coincidence degree in the at least one confusion knowledge point pair. According to the keyword coincidence degrees of the first M question pairs in the confusion knowledge point pair, the at least one confusion knowledge point pair is screened, which can avoid the question pairs with low keyword coincidence degrees and more accurately screen the confusion knowledge point pairs containing similar question pairs. Optionally, determining the expansion degree of the confusion knowledge point pair according to the keyword coincidence degrees of the first M question pairs comprises: calculating the average value of the keyword coincidence degrees of the first M question pairs as the expansion coincidence degree of the confusion knowledge point pair. Alternatively, determining the expansion degree of the confusion knowledge point pair according to the keyword coincidence degrees of the first M question pairs comprises: calculating the sum of the keyword coincidence degrees of the first M question pairs as the expansion coincidence degree of the confusion knowledge point pair. Of course, this is only an example.

[0039] Optionally, in the second example, after determining the at least one confusion knowledge point pair from the at least one candidate knowledge point pair according to the semantic similarity, the method further comprises: filtering the question pairs with the keyword coincidence degrees less than or equal to a preset threshold in the confusion knowledge point pair. By further filtering the question pairs according to the keyword coincidence degrees of the question pairs, the question pairs that are easy to confuse can be determined, and the accuracy is higher.

[0040] Based on the confusion knowledge point detection method described in the above embodiments, the application scenario of the knowledge base is described here after the confusion knowledge point pair is determined. After the confusion knowledge point pair is determined, the neural network model corresponding to the knowledge base can be further trained for the confusion knowledge point pair, so that the neural network model can distinguish the knowledge points in the confusion knowledge point pair. As shown in Figure 4 , the application scenario of the knowledge base provided by the first embodiment of the present application is shown in Figure 4 . Figure 4The terminal device 401, the cloud 402 and the user 403 are shown, wherein the terminal device 401 can be a smart speaker, a smart phone, a tablet computer, a notebook computer, etc. The terminal device 401 can access a network, connect with the cloud 402 through the network and perform data interaction. In the present application, the network includes a local area network (LAN), a wide area network (WAN), a mobile communication network, such as the World Wide Web (WWW), a Long Term Evolution (LTE) network, a 2G network, a 3G network, a 5G network, etc. Of course, this is only an exemplary illustration and does not represent a limitation of the present application. The cloud 402 can include various devices connected through the network, such as servers, relay devices, device-to-device (D2D) devices, etc.

[0041] The cloud 402 contains a knowledge base and a neural network model corresponding to the knowledge base. The user 403 can upload information input by the user to the cloud 402 through the terminal device 401, which can include text, images, videos or voice, etc. The terminal device 401 transmits the information input by the user 403 to the cloud 402, the cloud 402 identifies the information input by the user 403 using the neural network model, and determines a target knowledge point corresponding to the information input by the user 403 in the knowledge base according to the identification result, and transmits the target knowledge point to the terminal device 401, and the terminal device 401 displays the target knowledge point to the user. The target knowledge point can include at least one question and the corresponding answer. The user inputs information to the terminal device 401, and the terminal device 401 interacts with the cloud 402 to push the question and the answer related to the input information to the user, and more intelligently solve the user's doubts. Taking the use of an online shopping platform by the user as an example, the user can upload multimedia information (i.e. information input by the user) such as images, text, videos or voice to the online shopping platform through the terminal device 401, the cloud 402 identifies the multimedia information, determines the corresponding target knowledge point in the knowledge base, and returns the target knowledge point. For example, the user uploads an image of a commodity, and the target knowledge point can include a question such as "What is the price of the commodity?" and an answer of the price of the commodity. Of course, this is only an exemplary illustration.

[0042] The method for detecting confusing knowledge points provided in the embodiments of the present application determines a candidate set of confusing knowledge points according to the keyword coincidence degree between multiple knowledge points in a knowledge base; calculates the semantic similarity between a first knowledge point and a second knowledge point in the candidate set of confusing knowledge points; and determines the first knowledge point and the second knowledge point as confusing knowledge points if the semantic similarity is higher than a preset threshold similarity. The keyword coincidence degree can be used to quickly screen out the candidate set of confusing knowledge points, and then the semantic similarity can be used to more accurately identify the confusing knowledge points, thus balancing the identification accuracy and speed.

[0043] Embodiment Two

[0044] Based on the method for detecting confusing knowledge points described in Embodiment One, Embodiment Two of the present application provides a method for detecting confusing knowledge points, which is further described in detail with reference to Figure 5 Figure 5 A flowchart of the method for detecting confusing knowledge points provided in Embodiment Two of the present application is shown in the figure, which includes the following steps:

[0045] Step 501: Determine the keyword coincidence degree of at least one knowledge point pair according to the keywords of at least two knowledge points in a knowledge base.

[0046] The at least two knowledge points form at least one knowledge point pair.

[0047] Step 502: Determine at least one candidate knowledge point pair in the at least one knowledge point pair according to the keyword coincidence degree of the at least one knowledge point pair.

[0048] Step 503: Determine the semantic similarity of each candidate knowledge point pair in the at least one candidate knowledge point pair.

[0049] Step 504: Calculate the similarity of two questions in different knowledge points in the candidate knowledge point pair, to obtain the similarity of at least one question pair.

[0050] Step 505: Determine the semantic similarity of the candidate knowledge point pair according to the similarity of the at least one question pair.

[0051] Step 506: Determine at least one confusing knowledge point pair in the at least one candidate knowledge point pair according to the semantic similarity.

[0052] Step 507: Determine the keyword coincidence degree of the first M question pairs in the at least one question pair in descending order of similarity.

[0053] Step 508: Screen the at least one confusing knowledge point pair and the question pairs contained in the confusing knowledge point pair according to the keyword coincidence degree of the first M question pairs in each confusing knowledge point pair.

[0054] ​The steps 501-508 are explained in detail in Embodiment One and will not be repeated here. Based on the detection method of confusing knowledge points described above, a specific application scenario is listed here for illustration, such as Figure 6 as shown in Figure 6 FIG. 2, which is a schematic diagram of an application scenario of the detection method of confusing knowledge points provided in Embodiment Two of the present application, Figure 6 The electronic device 101 shown in Figure 1 may be the same as the electronic device shown in FIG. 1. The electronic device 101 first screens at least one knowledge point pair by using the keyword coincidence degree to determine at least one candidate knowledge point pair, and then further screens the at least one candidate knowledge point pair by using the semantic similarity to obtain at least one confusing knowledge point pair. Then, the electronic device 101 screens the confusing knowledge point pair and the question pair contained therein by using the keyword coincidence degree to obtain a confusing knowledge point pair and a question pair with higher accuracy. Through the three-step strategy of “keyword-semantic-keyword”, the keyword information and the semantic information are well integrated, and the accuracy of determining the confusing knowledge point pair is effectively improved.

[0055] The detection method of confusing knowledge points provided in the embodiments of the present application determines a confusing knowledge point candidate set according to the keyword coincidence degree between multiple knowledge points in a knowledge base; calculates the semantic similarity between a first knowledge point and a second knowledge point in the confusing knowledge point candidate set; and determines the first knowledge point and the second knowledge point as confusing knowledge points if the semantic similarity is higher than a preset threshold similarity. The keyword coincidence degree can be used to quickly screen out the confusing knowledge point candidate set, and then the semantic similarity can be used to more accurately identify the confusing knowledge points, taking into account the recognition accuracy and recognition speed.

[0056] Embodiment Three

[0057] Based on the methods described in Embodiments One and Two, Embodiment Three of the present application provides a detection device of confusing knowledge points for executing the methods described in Embodiments One and Two. Referring to FIG. 7, the detection device 70 of confusing knowledge points includes: Figure 7

[0058] A keyword module 701 is configured to determine a confusing knowledge point candidate set according to the keyword coincidence degree between multiple knowledge points in a knowledge base.

[0059] A semantic module 702 is configured to calculate the semantic similarity between a first knowledge point and a second knowledge point in the confusing knowledge point candidate set.

[0060] A confusing module 703 is configured to determine the first knowledge point and the second knowledge point as confusing knowledge points when the semantic similarity is higher than a preset threshold similarity.

[0061] ​Optionally, in an embodiment of the present application, the keyword module 701 is configured to determine keywords corresponding to each knowledge point; and perform matching using the keywords to obtain at least two knowledge points with keyword coincidence degrees higher than a threshold preset coincidence degree, to form a candidate set of confused knowledge points.

[0062] Optionally, in an embodiment of the present application, the keyword module 701 is configured to calculate keywords corresponding to each knowledge point using a term frequency inverse document frequency (TFIDF) method.

[0063] Optionally, in an embodiment of the present application, the knowledge points in the knowledge base include a plurality of questions and answers corresponding to the questions; and the semantic module 702 is configured to calculate two-by-two semantic similarities between the plurality of questions corresponding to the first knowledge point and the plurality of questions corresponding to the second knowledge point, and determine a semantic similarity between the first knowledge point and the second knowledge point according to the two-by-two semantic similarities.

[0064] Optionally, in an embodiment of the present application, the semantic module 702 is configured to calculate an average similarity of the first N question pairs in descending order of similarity, and determine the semantic similarity between the first knowledge point and the second knowledge point as the average similarity of the first N question pairs, where N is an integer greater than 0.

[0065] Optionally, in an embodiment of the present application, the semantic module 702 is configured to determine keyword coincidence degrees of the first M question pairs in descending order of similarity among the plurality of question pairs formed by the plurality of questions corresponding to the first knowledge point and the plurality of questions corresponding to the second knowledge point, and determine an extended coincidence degree between the first knowledge point and the second knowledge point according to the keyword coincidence degrees of the first M question pairs, where M is an integer greater than 0; and if the extended coincidence degree is greater than a preset extended coincidence degree, determine that the confused knowledge point formed by the first knowledge point and the second knowledge point is a high-quality confused knowledge point.

[0066] The detection apparatus for confused knowledge points provided in the embodiments of the present application determines a candidate set of confused knowledge points according to keyword coincidence degrees between a plurality of knowledge points in a knowledge base, calculates a semantic similarity between a first knowledge point and a second knowledge point in the candidate set of confused knowledge points, and determines that the first knowledge point and the second knowledge point are confused knowledge points if the semantic similarity is higher than a preset threshold similarity. The keyword coincidence degrees can be used to quickly screen out the candidate set of confused knowledge points, and the semantic similarity can be used to more accurately identify the confused knowledge points, thus balancing the identification accuracy and the identification speed.

[0067] Embodiment Four

[0068] Based on the methods described in Embodiments One and Two, Embodiment Four of the present application provides an electronic device configured to perform the methods described in Embodiments One and Two, as shown in FIG. 4. Figure 8 ​Figure 8 A structural schematic diagram of an electronic device is provided for Embodiment Four of the present application, and the specific implementation of the electronic device is not limited in the embodiments of the present application.

[0069] As shown in Figure 8 The electronic device 80 can include a processor 802, a communications interface 804, a memory 806, and a communications bus 808.

[0070] Among them:

[0071] The processor 802, the communications interface 804, and the memory 806 complete the communication with each other through the communications bus 808.

[0072]

[0073] The communications interface 804 is configured to communicate with other electronic devices, such as terminal devices or servers.

[0074] The processor 802 is configured to execute the program 810, and specifically can execute the related steps in the method embodiments.

[0075] Specifically, the program 810 can include program code, and the program code includes determination machine operation instructions.

[0076] The processor 802 can be a central processing unit CPU, or an application specific integrated circuit ASIC, or one or more integrated circuits configured to implement one or more embodiments of the present application. One or more processors included in the electronic device can be the same type of processor, such as one or more CPUs; or can be different types of processors, such as one or more CPUs and one or more ASICs.

[0077] The memory 806 is configured to store the program 810. The memory 806 can include a high-speed RAM memory, and can also include a non-volatile memory, for example, at least one disk memory.

[0078] The program 810 can be specifically used to enable the processor 802 to execute any one of the methods in Embodiment One and Embodiment Two.

[0079] ​The specific implementation of each step in the program 810 can refer to the corresponding description in the corresponding steps and units in the above-described detection method embodiments of the confusing knowledge points, which will not be described here. Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described devices and modules can refer to the corresponding process description in the foregoing method embodiments, which will not be described here.

[0080] The electronic device provided by the embodiment of the present application determines a candidate set of confusing knowledge points according to the keyword coincidence degree between multiple knowledge points in the knowledge base; calculates the semantic similarity between the first knowledge point and the second knowledge point in the candidate set of confusing knowledge points; and if the semantic similarity is higher than a preset threshold similarity, determines that the first knowledge point and the second knowledge point are confusing knowledge points. The keyword coincidence degree can be used to quickly screen out the candidate set of confusing knowledge points, and then the semantic similarity can be used to more accurately identify the confusing knowledge points, which takes into account the identification accuracy and identification speed.

[0081] Embodiment five

[0082] Based on the methods described in the above embodiment one and embodiment two, the embodiment five of the present application provides a computer storage medium having a determination program stored thereon, which is executed by a processor to implement the methods described in the embodiment one and the embodiment two.

[0083] The computer storage medium provided by the embodiment of the present application determines a candidate set of confusing knowledge points according to the keyword coincidence degree between multiple knowledge points in the knowledge base; calculates the semantic similarity between the first knowledge point and the second knowledge point in the candidate set of confusing knowledge points; and if the semantic similarity is higher than a preset threshold similarity, determines that the first knowledge point and the second knowledge point are confusing knowledge points. The keyword coincidence degree can be used to quickly screen out the candidate set of confusing knowledge points, and then the semantic similarity can be used to more accurately identify the confusing knowledge points, which takes into account the identification accuracy and identification speed.

[0084] It should be noted that, according to the needs of implementation, each component / step described in the embodiments of the present application can be split into more components / steps, or two or more components / steps or part of the operations of the components / steps can be combined into new components / steps, to achieve the purpose of the embodiments of the present application.

[0085] The method according to the embodiments of the present application described above can be implemented in hardware, firmware, or as software stored in a recording medium such as a CD ROM, a RAM, a floppy disk, a hard disk, or a magneto-optical disk, or as a program code to be stored in a remote recording medium or a non-transitory machine-readable medium and downloaded to a local recording medium, so that the method described herein can be processed by such software stored on a recording medium using a general purpose computer, a special purpose processor, or programmable or special hardware such as an ASIC or an FPGA. It can be appreciated that the computer, the processor, the microprocessor controller, or the programmable hardware includes a storage component (for example, a RAM, a ROM, a flash memory, etc.) that can store or receive software or program code, when the software or program code is accessed and executed by the computer, the processor, or the hardware, the detection method of the confusing knowledge points described herein is implemented. In addition, when a general purpose computer accesses the code for implementing the detection method of the confusing knowledge points described herein, the execution of the code will convert the general purpose computer into a special purpose computer for executing the detection method of the confusing knowledge points described herein.

[0086] Those skilled in the art can appreciate that the units and method steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. A skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of the present application.

[0087] The above embodiments are only used to illustrate but not limit the embodiments of the present application. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the embodiments of the present application, and all equivalent technical solutions belong to the scope of the embodiments of the present application. The patent protection scope of the embodiments of the present application should be defined by the claims.

Claims

1. A method for detecting confusing knowledge points, comprising: determining a confusing knowledge point candidate set according to keyword overlap degrees between a plurality of knowledge points in a knowledge base, the confusing knowledge point candidate set comprising at least one candidate knowledge point pair, the at least one candidate knowledge point pair comprising two knowledge points with a keyword overlap degree higher than a preset overlap degree; calculating a semantic similarity between a first knowledge point and a second knowledge point in the confusing knowledge point candidate set; determining the first knowledge point and the second knowledge point as confusing knowledge points if the semantic similarity is higher than a preset similarity; wherein the method further comprises: determining keyword overlap degrees of a first M question pairs in a plurality of question pairs composed of a plurality of questions corresponding to the first knowledge point and a plurality of questions corresponding to the second knowledge point in an order from high to low similarity, and determining an extended overlap degree between the first knowledge point and the second knowledge point according to the keyword overlap degrees of the first M question pairs, M being an integer greater than 0; determining a confusing knowledge point composed of the first knowledge point and the second knowledge point as a high-quality confusing knowledge point if the extended overlap degree is greater than a preset extended overlap degree.

2. The method of claim 1, wherein, The determining of the confusing knowledge point candidate set according to the keyword overlap degrees between the plurality of knowledge points in the knowledge base comprises: determining keywords corresponding to each knowledge point; using the keywords to match to obtain the confusing knowledge point candidate set comprising at least two knowledge points with a keyword overlap degree higher than a preset overlap degree.

3. The method of claim 2, wherein, The determining of the keywords corresponding to each knowledge point comprises using a term frequency-inverse document frequency (TFIDF) method to calculate the keywords corresponding to each knowledge point.

4. The method of claim 1, wherein, The knowledge points in the knowledge base comprise a plurality of questions and answers corresponding to the questions. The calculating of the semantic similarity between the first knowledge point and the second knowledge point in the confusing knowledge point candidate set comprises: calculating pairwise semantic similarities between a plurality of questions of the first knowledge point and a plurality of questions of the second knowledge point; determining the semantic similarity between the first knowledge point and the second knowledge point according to the pairwise semantic similarities.

5. The method of claim 4, wherein, The determining of the semantic similarity between the first knowledge point and the second knowledge point according to the pairwise semantic similarities comprises: calculating an average value of similarities of a first N question pairs in an order from high to low similarity, and determining the average value of the similarities as the semantic similarity between the first knowledge point and the second knowledge point, N being an integer greater than 0. 6.A device for detecting confusing knowledge points, comprising: a keyword module configured to determine a confusing knowledge point candidate set according to keyword overlap degrees between a plurality of knowledge points in a knowledge base, the confusing knowledge point candidate set comprising at least one candidate knowledge point pair, the at least one candidate knowledge point pair comprising two knowledge points with a keyword overlap degree higher than a preset overlap degree; a semantic module configured to calculate a semantic similarity between a first knowledge point and a second knowledge point in the confusing knowledge point candidate set; a confusing module configured to determine the first knowledge point and the second knowledge point as confusing knowledge points when the semantic similarity is higher than a preset threshold similarity; wherein the device further comprises: The determining module is configured to determine keyword coincidence degrees of the first M question pairs in order from large to small according to the similarity, and determine the expansion coincidence degree between the first knowledge point and the second knowledge point according to the keyword coincidence degrees of the first M question pairs, M being an integer greater than 0; if the expansion coincidence degree is greater than a preset expansion coincidence degree, the mixed knowledge point composed of the first knowledge point and the second knowledge point is determined as a high-quality mixed knowledge point.

7. An electronic device comprising: The processor, the memory, the communication interface and the communication bus complete communication with each other through the communication bus; The memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform the operation corresponding to the mixed knowledge point detection method in any one of claims 1-5. 8.A computer storage medium, having stored thereon a determination program, which, when executed by a processor, implements the mixed knowledge point detection method in any one of claims 1-5.

Images