Implicit evaluation object identification method, device, equipment and computer storage medium

Abstract

The application provides an implicit evaluation object identification method, device and equipment and a computer storage medium; the method comprises the following steps: receiving an evaluation information identification request, and obtaining commodity evaluation information associated with the evaluation information identification request; if the commodity evaluation information is first commodity evaluation information of an implicit evaluation object, inputting the first commodity evaluation information into a preset implicit evaluation object classifier; wherein the preset implicit evaluation object classifier is obtained through intermediate variable training of features, and the intermediate variable is a classification identification result of a feature classifier; processing the first commodity evaluation information through the implicit evaluation object classifier to obtain an evaluation object and sentiment information; in the application, the commodity evaluation information of the implicit evaluation object is classified and identified through the implicit evaluation object classifier, effective identification of the commodity evaluation information corresponding to the implicit evaluation object is realized, and the comprehensiveness and accuracy of the classification identification of the commodity evaluation information are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data analysis, in particular to an implicit evaluation object identification method, device and equipment and computer storage medium. BACKGROUND

[0002] With the development of Internet e-commerce, online shopping has become a normal part of people's lives. After online shopping, the purchased user will evaluate the purchased goods to facilitate the selection of the most suitable goods for the later buyers according to the goods evaluation information.

[0003] Analyzing and researching the goods evaluation information has important application value and economic value. For example, customers can choose suitable goods according to the goods evaluation information. On the other hand, manufacturers can understand the defects of their goods according to the goods evaluation information to update or improve the goods. However, there are thousands of goods evaluation information for a product, and users cannot browse all the goods evaluation information. E-commerce platforms summarize thousands of goods evaluation information and present it to users comprehensively. The summary of goods evaluation information includes three aspects: evaluation person identification, evaluation object identification and sentiment judgment. Generally, the evaluation person is not concerned during the summary of goods evaluation information, so the step of evaluation person identification is usually omitted. The evaluation object can be divided into explicit evaluation object and implicit evaluation object. In the existing evaluation object identification technology, the explicit evaluation object is usually identified, and the implicit evaluation object is ignored. After the evaluation object is identified, the sentiment of the evaluation object is usually judged, that is, whether the evaluation person is positive, negative or neutral.

[0004] Since the current goods evaluation information analysis is mainly for explicit evaluation objects, the implicit evaluation objects are ignored, which will lose a large amount of goods evaluation information and make the goods evaluation result inaccurate. SUMMARY

[0005] The present application provides an implicit evaluation object identification method, device, equipment and computer storage medium, which aims to solve the technical problem that the current goods evaluation data of implicit evaluation objects cannot be analyzed and summarized, resulting in incomplete goods evaluation information analysis and low accuracy of goods evaluation result identification.

[0006] In one aspect, the present application provides an implicit evaluation object identification method, which comprises the following steps:

[0007] receiving an evaluation information identification request and obtaining goods evaluation information associated with the evaluation information identification request;

[0008] If the commodity evaluation information is first commodity evaluation information of an implicit evaluation object, the first commodity evaluation information is input to a preset implicit evaluation object classifier; wherein the preset implicit evaluation object classifier is obtained through intermediate variable training of features, and the intermediate variable is a classification recognition result of a feature classifier.

[0009] The first commodity evaluation information is processed by the implicit evaluation object classifier to obtain an evaluation object and sentiment information.

[0010] In some embodiments of the present application, before the first commodity evaluation information of the implicit evaluation object is input to the preset implicit evaluation object classifier, the method comprises:

[0011] A model training request is received, a large amount of commodity evaluation information is obtained, the large amount of commodity evaluation information is processed, and an initial sample set is obtained;

[0012] According to the initial sample set and a preset dictionary, a feature sample set is constructed;

[0013] A preset proportion of feature samples is iteratively extracted from the feature sample set, a preset first classifier is trained through the feature samples, and a feature classifier is obtained;

[0014] The intermediate variable obtained by the classification recognition of the feature classifier is obtained, a preset second classifier is trained through the intermediate variable, and an implicit evaluation object classifier is obtained.

[0015] In some embodiments of the present application, the model training request is received, a large amount of commodity evaluation information is obtained, the large amount of commodity evaluation information is processed, and an initial sample set is obtained, which comprises:

[0016] A model training request is received, a large amount of commodity evaluation information is obtained, and each commodity evaluation information is added with a sentiment label;

[0017] The commodity evaluation information with the sentiment label is divided to form a first sample set of an implicit evaluation object and a second sample set of an explicit evaluation object;

[0018] Commodity evaluation information with each type of sentiment label is extracted from the first sample set and the second sample set respectively to form an initial sample set.

[0019] In some embodiments of the present application, the feature sample set is: D l = { ((MASK(x, v), IN(x, v)) | x ∈ D u}

[0020] The D irepresents a feature sample set, v represents a keyword in the preset dictionary, MASK represents that an element containing v in vector x is set to 0, IN represents judging whether v is contained in vector x, if v is contained, vector x is set to +1, if v is not contained, vector x is set to -1, D represents a preset dictionary, and u represents an initial sample set.

[0021] In some embodiments of the present application, the preset first classifier is:

[0022]

[0023] The w represents a dependent variable, and the w i represents a feature intermediate variable, L represents a loss function, w represents a dependent variable, and the w T represents a transpose of the w, and λ represents a preset constant.

[0024] In some embodiments of the present application, the feature intermediate variable obtained by classifying and identifying the feature classifier is used to train a preset second classifier to obtain an implicit evaluation object classifier, comprising:

[0025] The feature sample is processed by the feature classifier to obtain a feature parameter;

[0026] The feature parameter is spliced into a matrix, and singular value decomposition is performed on the matrix to obtain a feature intermediate variable;

[0027] A preset second classifier is trained by using the feature intermediate variable to obtain a trained classifier;

[0028] The classification and identification accuracy of the trained classifier is obtained, and the trained classifier with a classification and identification accuracy higher than a preset accuracy is used as a preset implicit evaluation object classifier.

[0029] In some embodiments of the present application, after receiving an evaluation information identification request and obtaining commodity evaluation information associated with the evaluation information identification request, the method comprises:

[0030] The commodity evaluation information is subjected to cluster analysis to determine the type of the commodity evaluation information;

[0031] If the commodity evaluation information is second commodity evaluation information of an explicit evaluation object, the second commodity evaluation information is input into a preset explicit evaluation object classifier;

[0032] The second commodity evaluation information is processed by the explicit evaluation object classifier to obtain sentiment information of the explicit evaluation object.

[0033] In one aspect, the present application provides an implicit evaluation object recognition device, comprising:

[0034] An information obtaining module is configured to receive an evaluation information recognition request and obtain product evaluation information associated with the evaluation information recognition request.

[0035] An information input module is configured to input first product evaluation information, which is implicit evaluation object, to a preset implicit evaluation object classifier if the product evaluation information is the first product evaluation information. The preset implicit evaluation object classifier is obtained through training of a feature intermediate variable, which is a classification recognition result of a feature classifier.

[0036] A processing output module is configured to process the first product evaluation information through the implicit evaluation object classifier to obtain evaluation object and sentiment information.

[0037] In another aspect, the present application also provides an implicit evaluation object recognition device, comprising:

[0038] One or more processors;

[0039] A memory; and

[0040] One or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processor to implement the implicit evaluation object recognition method.

[0041] In another aspect, the present application also provides a computer storage medium having a computer program stored thereon, wherein the computer program is loaded by a processor to execute steps in the implicit evaluation object recognition method.

[0042] In the technical solution of the present application: a product evaluation information recognition request is received, and product evaluation information associated with the product evaluation information recognition request is obtained; if the product evaluation information is first product evaluation information of an implicit evaluation object, the first product evaluation information is input to a preset implicit evaluation object classifier; the preset implicit evaluation object classifier is obtained through training of a feature intermediate variable, and the feature intermediate variable is a classification recognition result of a feature classifier; the implicit evaluation object classifier is used to process the first product evaluation information, and an evaluation object and sentiment information are obtained; in the present application, the preset implicit evaluation object classifier is used to process product evaluation information, obtain a feature intermediate variable, and connect implicit evaluation objects and explicit evaluation objects through the feature intermediate variable, so as to obtain an evaluation object and sentiment information of product evaluation information corresponding to an implicit evaluation object, and the technical solution of the present application realizes effective recognition of product evaluation information corresponding to an implicit evaluation object, and ensures the comprehensiveness and accuracy of product evaluation information classification recognition. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort.

[0044] Figure 1 is a scene schematic diagram of the implicit evaluation object recognition method provided by the present application embodiment;

[0045] Figure 2 is an embodiment flowchart of constructing an implicit evaluation object classifier in the implicit evaluation object recognition method in the present application embodiment;

[0046] Figure 3 is an embodiment scene schematic diagram of feature intermediate variable processing when constructing an implicit evaluation object classifier in the implicit evaluation object recognition method in the present application embodiment;

[0047] Figure 4 is an embodiment flowchart of the implicit evaluation object recognition method provided by the present application embodiment;

[0048] Figure 5 is another embodiment flowchart of the implicit evaluation object recognition method provided by the present application embodiment;

[0049] Figure 6 is an embodiment structure schematic diagram of the implicit evaluation object recognition device provided by the present application embodiment;

[0050] Figure 7is an embodiment structure schematic diagram of the implicit evaluation object recognition device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person skilled in the art without creative work fall within the scope of the present application.

[0052] In the description of the present application, it should be understood that the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0053] In the present application, the word "exemplary" is used to mean "serving as an example, instance, or illustration." Any implementation described as "exemplary" in the present application is not necessarily to be construed as preferred or advantageous over other implementations. The following description is presented to enable any person skilled in the art to make and use the present application. In the following description, for the purpose of explanation, details are set forth. It should be appreciated that one of ordinary skill in the art can realize and implement the present application without using these specific details. In other instances, well-known structures and processes are not described in detail in order to avoid obscuring the description of the present application. Therefore, the present application is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

[0054] The embodiment of the present application provides an implicit evaluation object recognition method, device, equipment and computer storage medium, which are described in detail below.

[0055] The implicit evaluation object recognition method in the embodiment of the present application is applied to an implicit evaluation object recognition device, the implicit evaluation object recognition device is arranged in an implicit evaluation object recognition equipment, the implicit evaluation object recognition equipment is provided with one or more processors, a memory and one or more application programs, wherein the one or more application programs are stored in the memory and are configured to be executed by the processor to realize the implicit evaluation object recognition method; the implicit evaluation object recognition equipment can be a terminal, such as a mobile phone or a tablet computer, and the implicit evaluation object recognition equipment can also be a server or a service cluster composed of multiple servers.

[0056] As Figure 1As shown, Figure 1 The scene of the implicit evaluation object identification method is a schematic diagram of an embodiment of the present application. In the embodiment of the present application, the implicit evaluation object identification scene includes an implicit evaluation object identification device 100 (an implicit evaluation object identification device 100 integrated with an implicit evaluation object identification apparatus). A computer storage medium corresponding to the implicit evaluation object identification is run in the implicit evaluation object identification device 100 to perform the steps of the implicit evaluation object identification.

[0057] It can be understood that, Figure 1 The implicit evaluation object identification device in the scene of the implicit evaluation object identification method, or the apparatus included in the implicit evaluation object identification device does not constitute a limitation on the embodiments of the present application, that is, the number and types of devices included in the scene of the implicit evaluation object identification method, or the number and types of apparatuses included in each device do not affect the overall implementation of the technical solutions in the embodiments of the present application, and can all be regarded as equivalent replacements or derivatives of the technical solutions claimed in the embodiments of the present application.

[0058] The implicit evaluation object identification device 100 in the embodiment of the present application is mainly used for receiving an evaluation information identification request, obtaining commodity evaluation information associated with the evaluation information identification request, inputting first commodity evaluation information of an implicit evaluation object to a preset implicit evaluation object classifier if the commodity evaluation information is the first commodity evaluation information of the implicit evaluation object, wherein the preset implicit evaluation object classifier is obtained through training of a feature intermediate variable, and the feature intermediate variable is a classification and identification result of a feature classifier. The evaluation object and the sentiment information are obtained by processing the first commodity evaluation information through the implicit evaluation object classifier.

[0059] The implicit evaluation object identification device 100 in the embodiment of the present application can be an independent implicit evaluation object identification device, or an implicit evaluation object identification device network or an implicit evaluation object identification device cluster composed of implicit evaluation object identification devices. For example, the implicit evaluation object identification device 100 described in the embodiment of the present application includes but is not limited to a computer, a network host, a single network implicit evaluation object identification device, a plurality of network implicit evaluation object identification device sets, or a cloud implicit evaluation object identification device composed of a plurality of implicit evaluation object identification devices. The cloud implicit evaluation object identification device is composed of a large number of computers or network implicit evaluation object identification devices based on cloud computing.

[0060] Those skilled in the art can understand that, Figure 1 The application environment shown in the embodiment of the present application is only one application scenario of the present application scheme, and does not constitute a limitation on the application scenarios of the present application scheme. Other application environments can also include more Figure 1more or less implicit evaluation object recognition devices, or implicit evaluation object recognition device network connection relationships, for example Figure 1 In the above embodiment, only 1 implicit evaluation object recognition device is shown, but it can be understood that the implicit evaluation object recognition method scenario can also include one or more other implicit evaluation object recognition devices, which are not limited here; the implicit evaluation object recognition device 100 can also include a memory for storing data.

[0061] In addition, in the implicit evaluation object recognition method scenario of the present application, the implicit evaluation object recognition device 100 can be provided with a display device, or the implicit evaluation object recognition device 100 is not provided with a display device and is communicatively connected with an external display device 200, and the display device 200 is used to output the results of the implicit evaluation object recognition method executed in the implicit evaluation object recognition device. The implicit evaluation object recognition device 100 can access a background database 300 (the background database can be a local memory of the implicit evaluation object recognition device, and the background database can also be set in the cloud), and the background database 300 stores information related to implicit evaluation object recognition, for example, the background database 300 stores product evaluation information.

[0062] It should be noted that Figure 1 The implicit evaluation object recognition method scenario shown in the above is only one example, and the implicit evaluation object recognition method scenario described in the embodiments of the present application is used to more clearly illustrate the technical solutions of the embodiments of the present application, and does not constitute a limitation on the technical solutions provided by the embodiments of the present application.

[0063] Based on the above implicit evaluation object recognition method scenario, an embodiment of an implicit evaluation object recognition method is provided, and the implicit evaluation object recognition method in the embodiment includes:

[0064] receiving an evaluation information recognition request and obtaining product evaluation information associated with the evaluation information recognition request;

[0065] If the product evaluation information is first product evaluation information of an implicit evaluation object, the first product evaluation information is input to a preset implicit evaluation object classifier; wherein the preset implicit evaluation object classifier is obtained by training a feature intermediate variable, and the feature intermediate variable is a classification recognition result of a feature classifier;

[0066] The first product evaluation information is processed by the implicit evaluation object classifier to obtain evaluation objects and sentiment information.

[0067] The implicit evaluation object recognition method in the embodiment is applied to an implicit evaluation object recognition device, and the type of the implicit evaluation object recognition device is not limited, for example, the implicit evaluation object recognition device can be a terminal or a server.

[0068] As Figure 2 shown, Figure 2 An embodiment flowchart of the implicit evaluation object classification method in the embodiment of the present application is shown.

[0069] Before the steps of the implicit evaluation object recognition method in the embodiment are executed, an implicit evaluation object classifier is preset in the implicit evaluation object recognition device. The implicit evaluation object classifier refers to a pre-set classification and recognition algorithm for the product evaluation information of the implicit evaluation object. The implicit evaluation object classifier construction in the implicit evaluation object recognition method in the embodiment includes the following steps 201-204.

[0070] 201, receiving a model training request, obtaining a large amount of product evaluation information, processing the large amount of product evaluation information, and obtaining an initial sample set.

[0071] The implicit evaluation object recognition device receives a model training request. The triggering mode of the model training request is not limited, i.e., the model training request can be manually triggered by a user, for example, the user presses the "training" button on the display interface of the implicit evaluation object recognition device to trigger the model training request. Alternatively, the model training request can also be automatically triggered by the implicit evaluation object recognition device, for example, the implicit evaluation object recognition device is preset to update the model every week, and the implicit evaluation object recognition device automatically triggers the model training request when it detects that the preset time point is reached.

[0072] The implicit evaluation object recognition device receives a model training request. The implicit evaluation object recognition device obtains a large amount of product evaluation information, adds emotional category labels to the large amount of product evaluation information, and extracts product evaluation information from product evaluation information of each type of emotional label to form an initial sample set. Specifically, step 201 includes:

[0073] (1) receiving a model training request, obtaining a large amount of product evaluation information, and adding emotional labels to each product evaluation information;

[0074] (2) dividing the product evaluation information with emotional labels to form a first sample set of implicit evaluation objects and a second sample set of explicit evaluation objects;

[0075] (3) extracting product evaluation information with each type of emotional label from the first sample set and the second sample set to form an initial sample set.

[0076] That is, the implicit evaluation object recognition device receives a model training request, the implicit evaluation object recognition device obtains a large amount of commodity evaluation information, and each of the commodity evaluation information is added with an emotional label; wherein the way in which the implicit evaluation object recognition device adds the emotional label to the commodity evaluation information is not limited in the embodiment, and specifically, implementation mode one: the emotional label set is arranged in the implicit evaluation object recognition device, the emotional label set includes different types of emotional labels, each emotional label includes words with the same emotion, the implicit evaluation object recognition device performs word segmentation processing on the commodity evaluation information to obtain evaluation keywords, the implicit evaluation object recognition device compares the evaluation keywords with the emotional words in each emotional label in the emotional label set to determine whether they match, so as to add the emotional label to the commodity evaluation information; implementation mode two: the commodity evaluation information is classified into different categories, and different categories of commodity evaluation information are added with corresponding emotional labels.

[0077] The implicit evaluation object recognition device divides the commodity evaluation information with the emotional label, that is, the implicit evaluation object recognition device determines whether the evaluation object in the commodity evaluation information is limited, the implicit evaluation object recognition device collects the commodity evaluation information without the limited evaluation object to form a first sample set of the implicit evaluation object, and the implicit evaluation object recognition device collects the commodity evaluation information with the limited evaluation object to form a second sample set of the explicit evaluation object.

[0078] The implicit evaluation object recognition device extracts commodity evaluation information with each type of emotional label from the first sample set and the second sample set to form an initial sample set, in the embodiment, the emotional label is added to the large amount of commodity evaluation information, and the first sample set and the second sample set are divided according to whether the evaluation object in the commodity evaluation information is limited, the implicit evaluation object recognition device extracts the commodity evaluation information corresponding to each type of emotional label from the first sample set and the second sample set respectively to form the initial sample set, which can effectively guarantee the diversity and comprehensiveness of the training sample to realize accurate model construction.

[0079] 202, according to the initial sample set and the preset dictionary, a feature sample set is constructed.

[0080] The implicit evaluation object recognition device compares the commodity evaluation information in the initial sample set with each keyword in the preset dictionary to obtain a word vector, and the word vector is collected to obtain a feature sample set. Wherein, the preset dictionary refers to a pre-set keyword set, the preset dictionary can be obtained by performing word segmentation processing on the evaluation information, and the preset dictionary can be a word vector set.

[0081] In the embodiment, the implicit evaluation object recognition device extracts features from the initial sample set according to the preset field to obtain a feature sample set, so as to train the classifier through the feature sample set, and specifically:

[0082] 203, iteratively extract a preset proportion of feature samples from the feature sample set, train a preset first classifier through the feature samples, and obtain a feature classifier.

[0083] The implicit evaluation object recognition device iteratively extracts a preset proportion of feature samples from the feature sample set, wherein the preset proportion can be set according to a specific scene, for example, the preset proportion is set to 0.1%, and the feature sample set is: D l = { ((MASK(x, v), IN(x, v)) | x e D u}; The D i indicates the feature sample set, the v indicates the keyword in the preset dictionary, the MASK indicates that the element containing v in the vector x is set to 0, the IN indicates that it is judged whether the vector x contains v, if the vector x contains v, the vector x is set to +1, if the vector x does not contain v, the vector x is set to -1, the D u indicates the initial sample set, as shown in Figure 3 , and Figure 3 is an embodiment scene diagram of processing a feature intermediate variable in the construction of an implicit evaluation object classifier in the implicit evaluation object recognition method in the embodiment of the application.

[0084] The implicit evaluation object recognition device trains a preset first classifier through the feature sample, and obtains a feature classifier. Wherein, the preset first classifier is:

[0085]

[0086] The w i indicates a feature intermediate variable, the L indicates a loss function, the w indicates a dependent variable, and the w T indicates the transpose of the w, and the λ indicates a preset constant.

[0087] In the embodiment, the preset first classifier is an initial feature extraction model, the implicit evaluation object is trained through the feature training sample, the feature classifier is obtained, the feature intermediate variable can be extracted through the feature classifier, the preset second classifier is trained through the feature intermediate variable, and the implicit evaluation object classifier is constructed. Specifically:

[0088] 204, obtain the feature intermediate variable obtained by classifying and recognizing the feature classifier, train a preset second classifier through the feature intermediate variable, and obtain an implicit evaluation object classifier.

[0089] The implicit evaluation object recognition device obtains the feature intermediate variable obtained by classifying and recognizing the feature classifier, the implicit evaluation object recognition device trains a preset second classifier through the feature intermediate variable, and obtains an implicit evaluation object classifier. Specifically, step 204 includes:

[0090] (1), processing the feature sample through the feature classifier to obtain a feature parameter;

[0091] (2), splicing the feature parameter to form a matrix, singular value decomposition is performed on the matrix to obtain a feature intermediate variable;

[0092] (3), training a preset second classifier through the feature intermediate variable to obtain a trained classifier;

[0093] (4), obtaining the classification recognition accuracy of the trained classifier, and taking the trained classifier with the classification recognition accuracy higher than a preset accuracy as a preset implicit evaluation object classifier.

[0094] That is, in the embodiment, the implicit evaluation object recognition device processes the feature sample through the feature classifier to obtain the feature parameter; the implicit evaluation object recognition device splices the feature parameter to form a matrix, and performs singular value decomposition on the matrix to obtain the feature intermediate variable. Specifically, the implicit evaluation object recognition device obtains the feature parameter w obtained by processing the feature classifier, and splices the feature parameter w to form a matrix W, and then performs singular value decomposition on the matrix W to obtain the feature intermediate variable wherein the feature intermediate variable is singular value decomposition of W, that is, UΣV T =SVD(W), θ is obtained from U.

[0095] The implicit evaluation object recognition device trains a preset second classifier through the feature intermediate variable to obtain an implicit evaluation object classifier, wherein the preset second classifier refers to a preset implicit evaluation object classification algorithm, and the algorithm of the preset second classifier is not limited.

[0096] Specifically, the implicit evaluation object recognition device trains the preset second classifier to obtain a trained classifier, the implicit evaluation object recognition device obtains the classification recognition accuracy of the trained classifier, and the implicit evaluation object recognition device compares the classification recognition accuracy with a preset accuracy, wherein the preset accuracy is set according to a specific scene, for example, the classification recognition accuracy is set to 98%, if the classification recognition accuracy is not higher than the preset accuracy, the implicit evaluation object recognition device determines that the trained classifier does not converge, and the implicit evaluation object recognition device iteratively trains the classifier; or, if the classification recognition accuracy is higher than the preset accuracy, the implicit evaluation object recognition device determines that the trained classifier converges, and the implicit evaluation object recognition device takes the trained classifier with the classification recognition accuracy higher than the preset accuracy as the preset implicit evaluation object classifier.

[0097] In this embodiment, the explicit evaluation object training set that has been labeled is used to select the feature intermediate variable through the mutual information between the evaluation object and the word. The feature intermediate variable, the first training set corresponding to the labeled explicit evaluation object, and the second training set of the unmarked implicit evaluation object, the feature classifier, and the implicit evaluation object classifier are used to effectively identify the implicit evaluation object through the implicit evaluation object classifier. In this embodiment, the implicit evaluation object classifier is effectively constructed, and the commodity evaluation information is analyzed through the implicit evaluation object classifier to realize comprehensive evaluation of data.

[0098] As shown in Figure 4 , the implicit evaluation object recognition method in the embodiment of the application is an embodiment flowchart, which includes the following steps 301-303. Figure 4

[0099] 301, receive an evaluation information recognition request, and obtain commodity evaluation information associated with the evaluation information recognition request.

[0100] The implicit evaluation object recognition device receives an evaluation information recognition request, wherein the trigger mode of the evaluation information recognition request is not limited, that is, the evaluation information recognition request can be triggered by the user, for example, the user's voice is: classify and identify the commodity evaluation information of the xxx commodity in March 2021, the implicit evaluation object recognition device receives the evaluation information recognition request, in addition, the evaluation information recognition request can also be triggered automatically by the implicit evaluation object recognition device, for example, the implicit evaluation object recognition device is pre-set to automatically trigger the evaluation information recognition request at dawn every day, and the implicit evaluation object recognition device automatically triggers the evaluation information recognition request at dawn every day.

[0101] After the implicit evaluation object recognition device receives the evaluation information recognition request, the implicit evaluation object recognition device obtains the commodity evaluation information associated with the evaluation information recognition request, wherein the category and content of the commodity evaluation information are not limited, for example, the commodity evaluation information can be that the appearance of xx is very good and the cost performance is very high, and it is recommended to buy.

[0102] 302, if the commodity evaluation information is the first commodity evaluation information of the implicit evaluation object, the first commodity evaluation information is input into the preset implicit evaluation object classifier; wherein the preset implicit evaluation object classifier is obtained by training the feature intermediate variable, and the feature intermediate variable is the classification recognition result of the feature classifier.

[0103] ​The implicit evaluation object recognition device identifies product evaluation information and determines whether the product evaluation information limits the evaluation object. The implicit evaluation object recognition device takes the product evaluation information without limiting the evaluation object as the first product evaluation information, and the implicit evaluation object recognition device takes the second product evaluation information that limits the evaluation object. For example, the first product evaluation information is: the size is very suitable; the second product evaluation information is: this cup is beautiful and the material is very good.

[0104] The implicit evaluation object recognition device determines the product evaluation information as the first product evaluation information of the implicit evaluation object, and then inputs the first product evaluation information into a preset implicit evaluation object classifier; wherein, the preset implicit evaluation object classifier is obtained by training through feature intermediate variables, and the feature intermediate variables are the classification and recognition results of the feature classifier.

[0105] 303. Process the first product evaluation information through the implicit evaluation object classifier to obtain the evaluation object and sentiment information.

[0106] The implicit evaluation object recognition device processes the evaluation information of the first product through an implicit evaluation object classifier to obtain the evaluation object and sentiment information. That is, the implicit evaluation object recognition device processes the evaluation information of the first product through a feature classification model to obtain intermediate feature variables. The implicit evaluation object recognition device inputs the intermediate feature variables into the implicit evaluation object classifier to obtain the evaluation object and sentiment information corresponding to the evaluation information of the first product.

[0107] In this embodiment, an evaluation information identification request is received, and the product evaluation information associated with the evaluation information identification request is obtained. If the product evaluation information is the first product evaluation information of an implicit evaluation object, the first product evaluation information is input to a preset implicit evaluation object classifier. The preset implicit evaluation object classifier is obtained through training with feature intermediate variables, where the feature intermediate variables are the classification and identification results of the feature classifier. The first product evaluation information is processed by the implicit evaluation object classifier to obtain evaluation object and sentiment information. In this embodiment, a preset implicit evaluation object classifier is used. The implicit evaluation object classifier can process the product evaluation information to obtain feature intermediate variables. By connecting the implicit evaluation object and the explicit evaluation object through the feature intermediate variables, the evaluation object and sentiment information of the product evaluation information corresponding to the implicit evaluation object are obtained. This technical solution achieves effective identification of the product evaluation information corresponding to the implicit evaluation object, ensuring the comprehensiveness and accuracy of product evaluation information classification and identification.

[0108] like Figure 5 As shown, Figure 5 This is a schematic flowchart of another embodiment of the implicit evaluation object identification method provided in this application.

[0109] In some embodiments of this application, the steps for constructing an image segmentation model are specifically described, including the following steps 401 to 403:

[0110] 401. Perform cluster analysis on the product review information to determine the type of product review information.

[0111] The implicit review object recognition device performs cluster analysis on product review information to determine the type of product review information. In this embodiment, the implicit review object recognition device confirms the type of product review information and can determine the processing method for the product review information based on the type of product review information. Specifically:

[0112] 402. If the product evaluation information is the second product evaluation information of an explicit evaluation object, then the second product evaluation information is input into the preset explicit evaluation object classifier.

[0113] If the product evaluation information is the second product evaluation information of the explicit evaluation object, the implicit evaluation object recognition device will input the second product evaluation information into the preset explicit evaluation object classifier, wherein the explicit evaluation object classifier is a pre-trained classification model.

[0114] 403. Process the second product evaluation information through the explicit evaluation object classifier to obtain the sentiment information of the explicit evaluation object.

[0115] The implicit evaluation object recognition device processes the second product evaluation information through an explicit evaluation object classifier to obtain the emotional information of the explicit evaluation object. In this embodiment, the product evaluation information of the explicit evaluation object can be analyzed, ensuring the comprehensiveness of the product evaluation information processing.

[0116] To better implement the implicit evaluation object identification method in the embodiments of this application, an implicit evaluation object identification device is also provided in the embodiments of this application, such as... Figure 6 As shown, Figure 6 This is a schematic diagram of an embodiment of an implicit evaluation object identification device, which includes the following modules 501-503:

[0117] The information acquisition module 501 is used to receive an evaluation information identification request and acquire the product evaluation information associated with the evaluation information identification request;

[0118] The information input module 502 is used to input the first product evaluation information into a preset implicit evaluation object classifier if the product evaluation information is the first product evaluation information of the implicit evaluation object; wherein the preset implicit evaluation object classifier is obtained by training through feature intermediate variables, and the feature intermediate variables are the classification and recognition results of the feature classifier.

[0119] The processing output module 503 is configured to process the first commodity evaluation information by using the implicit evaluation object classifier to obtain an evaluation object and sentiment information.

[0120] In some embodiments of the present application, the implicit evaluation object recognition device comprises:

[0121] The model training request is received, and a large amount of commodity evaluation information is obtained. The large amount of commodity evaluation information is processed to obtain an initial sample set.

[0122] According to the initial sample set and a preset dictionary, a feature sample set is constructed.

[0123] A preset proportion of feature samples are iteratively extracted from the feature sample set, and a preset first classifier is trained by using the feature samples to obtain a feature classifier.

[0124] A feature intermediate variable obtained by classification and recognition of the feature classifier is obtained, and a preset second classifier is trained by using the feature intermediate variable to obtain an implicit evaluation object classifier.

[0125] In some embodiments of the present application, the implicit evaluation object recognition device executes the receiving of the model training request, the obtaining of a large amount of commodity evaluation information, the processing of the large amount of commodity evaluation information, and the obtaining of an initial sample set, which comprises:

[0126] The model training request is received, and a large amount of commodity evaluation information is obtained. Each of the commodity evaluation information is added with a sentiment label.

[0127] The commodity evaluation information with the sentiment label is divided to form a first sample set of implicit evaluation objects and a second sample set of explicit evaluation objects.

[0128] Commodity evaluation information with each type of sentiment label is extracted from the first sample set and the second sample set to form an initial sample set.

[0129] In some embodiments of the present application, the feature sample set in the implicit evaluation object recognition device is D l = { ((MASK(x, v), IN(x, v)) | x e D u}

[0130] D i represents a feature sample set, v represents a keyword in the preset dictionary, MASK represents that an element containing v in a vector x is set to 0, IN represents that it is judged whether v is contained in the vector x, if v is contained, the vector x is set to +1, and if v is not contained, the vector x is set to -1, and D u represents an initial sample set.

[0131] In some embodiments of the present application, the preset first classifier in the implicit evaluation object recognition device is:

[0132]

[0133] The w i represents a characteristic intermediate variable, the L represents a loss function, the w represents a dependent variable, and the w T represents the transpose of the w, and the λ represents a preset constant.

[0134] In some embodiments of the present application, the implicit evaluation object recognition device performs the following steps:

[0135] The feature sample is processed by the feature classifier to obtain a feature parameter;

[0136] The feature parameter is spliced into a matrix, and singular value decomposition is performed on the matrix to obtain a feature intermediate variable;

[0137] A preset second classifier is trained by the feature intermediate variable to obtain a trained classifier;

[0138] The classification recognition accuracy of the trained classifier is obtained, and the trained classifier with a classification recognition accuracy higher than a preset accuracy is taken as a preset implicit evaluation object classifier.

[0139] In some embodiments of the present application, the implicit evaluation object recognition device further comprises:

[0140] The commodity evaluation information is subjected to cluster analysis to determine the type of the commodity evaluation information;

[0141] If the commodity evaluation information is second commodity evaluation information of an explicit evaluation object, the second commodity evaluation information is input into a preset explicit evaluation object classifier;

[0142] The second commodity evaluation information is processed by the explicit evaluation object classifier to obtain the sentiment information of the explicit evaluation object.

[0143] The implicit evaluation object recognition device in the embodiment receives an evaluation information recognition request, acquires commodity evaluation information associated with the evaluation information recognition request, inputs first commodity evaluation information of an implicit evaluation object into a preset implicit evaluation object classifier if the commodity evaluation information is the first commodity evaluation information of the implicit evaluation object, wherein the preset implicit evaluation object classifier is obtained through training of a feature intermediate variable, the feature intermediate variable is a classification recognition result of a feature classifier, and the first commodity evaluation information is processed by the implicit evaluation object classifier to obtain an evaluation object and sentiment information. In the technical solution of the application, the preset implicit evaluation object classifier can process commodity evaluation information to obtain a feature intermediate variable, and the implicit evaluation object is connected to an explicit evaluation object through the feature intermediate variable to obtain the evaluation object and sentiment information of the commodity evaluation information corresponding to the implicit evaluation object. The technical solution of the application realizes effective recognition of the commodity evaluation information corresponding to the implicit evaluation object, and guarantees the comprehensiveness and accuracy of classification recognition of the commodity evaluation information.

[0144] The embodiment of the application further provides an implicit evaluation object recognition device, as shown in Figure 7 Figure 7 is an embodiment structure schematic diagram of the implicit evaluation object recognition device provided in the embodiment of the application.

[0145] The implicit evaluation object recognition device integrates any one of the implicit evaluation object recognition devices provided in the embodiment of the application, and the implicit evaluation object recognition device comprises:

[0146] one or more processors;

[0147] a memory; and

[0148] one or more application programs, wherein the one or more application programs are stored in the memory and are configured to execute steps in the implicit evaluation object recognition method in any one of the embodiment of the implicit evaluation object recognition method.

[0149] Specifically, the implicit evaluation object recognition device can comprise a processor 601 with one or more processing cores, a memory 602 with one or more computer storage media, a power supply 603, an input unit 604 and the like. Those skilled in the art can understand that the structure of the implicit evaluation object recognition device shown in Figure 7 the embodiment of the application does not constitute a limitation on the implicit evaluation object recognition device, and can comprise more or fewer components than those shown in the figure, or combine certain components, or different arrangement of components. Among them:

[0150] ​The processor 601 is the control center of the implicit evaluation object recognition device, connects each part of the implicit evaluation object recognition device through various interfaces and lines, executes various functions of the implicit evaluation object recognition device and processes data by running or executing software programs and / or modules stored in the memory 602 and calling data stored in the memory 602, and thus monitors the whole implicit evaluation object recognition device. Optionally, the processor 601 can include one or more processing cores; preferably, the processor 601 can integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface and application program, etc., and the modem processor mainly processes wireless communication. It can be understood that the above-mentioned modem processor can also not be integrated into the processor 601.

[0151] The memory 602 can be used to store software programs and modules, and the processor 601 executes various functions and data processing by running the software programs and modules stored in the memory 602. The memory 602 can mainly include a program storage area and a data storage area, wherein the program storage area can store the operating system, at least one application program required by the function (such as sound playing function, image playing function, etc.), etc.; the data storage area can store data created according to the use of the implicit evaluation object recognition device, etc. In addition, the memory 602 can include a high-speed random access memory, and can also include a non-volatile memory, for example, at least one magnetic disk storage device, flash memory device, or other volatile solid-state memory device. Accordingly, the memory 602 can also include a memory controller to provide the processor 601 with access to the memory 602.

[0152] The implicit evaluation object recognition device further includes a power supply 603 for supplying power to each component, and preferably, the power supply 603 can be logically connected with the processor 601 through a power management system, so as to realize the functions of managing charging, discharging and power consumption management, etc. through the power management system. The power supply 603 can also include one or more than one direct current or alternating current power supply, a recharging system, a power failure detection circuit, a power converter or inverter, a power state indicator and any other components.

[0153] The implicit evaluation object recognition device can also include an input unit 604, which can be used to receive input digital or character information, and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.

[0154] Although not shown, the implicit evaluation object recognition device can also include a display unit or the like, which will not be described here. In particular, in the present embodiment, the processor 601 in the implicit evaluation object recognition device will load the executable file corresponding to the process of one or more application programs into the memory 602 according to the following instructions, and run the application program stored in the memory 602 by the processor 601, thereby realizing various functions, as follows:

[0155] receiving an evaluation information recognition request, and obtaining product evaluation information associated with the evaluation information recognition request;

[0156] If the product evaluation information is first product evaluation information of an implicit evaluation object, the first product evaluation information is input to a preset implicit evaluation object classifier; wherein the preset implicit evaluation object classifier is obtained by training a feature intermediate variable, and the feature intermediate variable is a classification recognition result of a feature classifier;

[0157] The first product evaluation information is processed by the implicit evaluation object classifier to obtain evaluation objects and sentiment information.

[0158] Those skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructions, or by related hardware controlled by instructions, which can be stored in a computer storage medium and loaded and executed by a processor.

[0159] To this end, the embodiments of the present application provide a computer storage medium, which can include a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc. A computer program is stored thereon, and the computer program is loaded by a processor to execute the steps in any of the implicit evaluation object recognition methods provided by the embodiments of the present application. For example, the computer program loaded by the processor can execute the following steps:

[0160] receiving an evaluation information recognition request, and obtaining product evaluation information associated with the evaluation information recognition request;

[0161] If the product evaluation information is first product evaluation information of an implicit evaluation object, the first product evaluation information is input to a preset implicit evaluation object classifier; wherein the preset implicit evaluation object classifier is obtained by training a feature intermediate variable, and the feature intermediate variable is a classification recognition result of a feature classifier;

[0162] The first product evaluation information is processed by the implicit evaluation object classifier to obtain evaluation objects and sentiment information.

[0163] In the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the detailed description of other embodiments above, which will not be repeated here.

[0164] In the implementation, each unit or structure above can be implemented as an independent entity, or can be combined as the same or several entities, and the specific implementation of each unit or structure above can be referred to the method embodiments above, which will not be repeated here.

[0165] The specific implementation of each operation above can be referred to the embodiments above, which will not be repeated here.

[0166] The above has carried out the detailed introduction to the implicit evaluation object recognition method provided by the embodiment of the application, the principle and implementation mode of the application are described in this paper by applying specific examples, the above embodiment is only used to help understanding the method and core idea of the application; at the same time, for those skilled in the art, according to the idea of the application, the specific implementation mode and application range will have changes, and the above is not understood as the limitation of the application.

Claims

1. An implicit evaluation object recognition method characterized by, The method comprises: receiving an evaluation information identification request, and obtaining commodity evaluation information associated with the evaluation information identification request; if the commodity evaluation information is first commodity evaluation information of an implicit evaluation object, inputting the first commodity evaluation information into a preset implicit evaluation object classifier; wherein the preset implicit evaluation object classifier is obtained through intermediate variable training of features, and the intermediate variable of the features is a classification and identification result of a feature classifier; processing the first commodity evaluation information through the implicit evaluation object classifier to obtain an evaluation object and sentiment information; wherein, before the step of, if the commodity evaluation information is first commodity evaluation information of an implicit evaluation object, inputting the first commodity evaluation information into a preset implicit evaluation object classifier, the method comprises: receiving a model training request, obtaining a large amount of commodity evaluation information, processing the large amount of commodity evaluation information to obtain an initial sample set; constructing a feature sample set according to the initial sample set and a preset dictionary; iteratively extracting a preset proportion of feature samples from the feature sample set, training a preset first classifier through the feature samples to obtain a feature classifier; processing the feature samples through the feature classifier to obtain feature parameters; splicing the feature parameters to form a matrix, singular value decomposition is performed on the matrix to obtain an intermediate variable of the features; training a preset second classifier through the intermediate variable of the features to obtain a trained classifier; obtaining the classification and identification accuracy of the trained classifier, and taking the trained classifier with a classification and identification accuracy higher than a preset accuracy as a preset implicit evaluation object classifier.

2. The implicit evaluation target recognition method according to claim 1, characterized by, The step of, receiving a model training request, obtaining a large amount of commodity evaluation information, processing the large amount of commodity evaluation information to obtain an initial sample set, comprises: receiving a model training request, obtaining a large amount of commodity evaluation information, and adding sentiment labels to each of the commodity evaluation information; dividing the commodity evaluation information with sentiment labels to form a first sample set of implicit evaluation objects and a second sample set of explicit evaluation objects; extracting commodity evaluation information with various types of sentiment labels from the first sample set and the second sample set to form an initial sample set.

3. The implicit evaluation target recognition method according to claim 1, characterized by, The feature sample set is: The D l represents a feature sample set, v represents a keyword in the preset dictionary, MASK represents setting elements corresponding to v in vector x to 0, IN represents judging whether v is contained in vector x, if v is contained, vector x is set to +1, if v is not contained, vector x is set to -1, and D u represents an initial sample set.

4. The method according to any one of claims 1 to 3, wherein After the step of, receiving an evaluation information identification request, and obtaining commodity evaluation information associated with the evaluation information identification request, the method comprises: performing cluster analysis on the commodity evaluation information to determine the type of the commodity evaluation information; if the commodity evaluation information is second commodity evaluation information of an explicit evaluation object, inputting the second commodity evaluation information into a preset explicit evaluation object classifier; processing the second commodity evaluation information through the explicit evaluation object classifier to obtain sentiment information of the explicit evaluation object.

5. An implicit evaluation object recognition apparatus characterized by comprising: The implicit evaluation object identification device comprises: an information obtaining module configured to receive an evaluation information identification request and obtain commodity evaluation information associated with the evaluation information identification request; The information input module is configured to, if the commodity evaluation information is first commodity evaluation information of an implicit evaluation object, input the first commodity evaluation information to a preset implicit evaluation object classifier. The preset implicit evaluation object classifier is obtained through intermediate variable training of features, and the intermediate variable is a classification and recognition result of a feature classifier. The processing output module is configured to process the first commodity evaluation information through the implicit evaluation object classifier to obtain an evaluation object and sentiment information. Before the implicit evaluation object recognition device executes the first commodity evaluation information of the implicit evaluation object, the device is further configured to: receive a model training request, obtain a large amount of commodity evaluation information, process the large amount of commodity evaluation information to obtain an initial sample set; construct a feature sample set according to the initial sample set and a preset dictionary; extract a preset proportion of feature samples from the feature sample set, train a preset first classifier through the feature samples, and obtain a feature classifier; process the feature samples through the feature classifier to obtain feature parameters; splice the feature parameters to form a matrix, perform singular value decomposition on the matrix to obtain an intermediate variable of features, and train a preset second classifier through the intermediate variable of features to obtain a trained classifier; obtain a classification and recognition accuracy of the trained classifier, and use a trained classifier with a classification and recognition accuracy higher than a preset accuracy as a preset implicit evaluation object classifier. The implicit evaluation object recognition device comprises:

6. An implicit evaluation object recognition apparatus characterized by comprising: one or more processors; a memory; and one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processor to implement the implicit evaluation object recognition method of any one of claims 1 to 4. A computer program is stored thereon, and the computer program is loaded by a processor to execute the steps in the implicit evaluation object recognition method of any one of claims 1 to 4.

7. A computer storage medium, characterized in that ​

Images