CN115168726A - Model training method and information recommendation method - Google Patents
Model training method and information recommendation method Download PDFInfo
- Publication number
- CN115168726A CN115168726A CN202210851513.8A CN202210851513A CN115168726A CN 115168726 A CN115168726 A CN 115168726A CN 202210851513 A CN202210851513 A CN 202210851513A CN 115168726 A CN115168726 A CN 115168726A
- Authority
- CN
- China
- Prior art keywords
- evaluation
- dimension
- evaluation dimension
- feature data
- knowledge graph
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Economics (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Animal Behavior & Ethology (AREA)
- Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The specification discloses a model training method and an information recommendation method, wherein a server acquires comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension. Secondly, inputting the comment data and the knowledge graph corresponding to the comment dimension into a prediction model to be trained aiming at each comment dimension so as to determine comment characteristic data corresponding to the comment data and dimension characteristic data corresponding to the comment dimension. And then, determining an evaluation result of the user for evaluating the specified object by the evaluation dimension according to the dimension characteristic data corresponding to the evaluation dimension and the evaluation characteristic data. And finally, training the prediction model by taking the deviation between the evaluation result of evaluating the specified object by minimizing each evaluation dimension and the label corresponding to the comment data as an optimization target.
Description
Technical Field
The present disclosure relates to the field of computer technologies, and in particular, to a model training method and an information recommendation method.
Background
With the rapid development of the internet technology, the comment information of the commodity has a great influence on the purchasing behavior of the user in the process of purchasing the commodity.
At present, a server inputs user comment data into a prediction model, and can only predict whether the overall evaluation of the user comment data is good, or whether the evaluation of a certain aspect in the user comment data is good. And the evaluation in a single aspect can not enable the user to comprehensively know the advantages and disadvantages of the commodities, so that reasonable purchasing behavior is made, and the purchasing experience of the user is poor.
Therefore, how to effectively improve the purchasing experience of the user is a problem to be solved urgently.
Disclosure of Invention
The present specification provides a method of model training to partially solve the above problems of the prior art.
The technical scheme adopted by the specification is as follows:
the present specification provides a method of model training, comprising:
obtaining comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension;
for each evaluation dimension, inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a prediction model to be trained so as to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension;
determining an evaluation result of the user for evaluating the specified object by the evaluation dimension according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data;
and training the prediction model by taking the deviation between the evaluation result of evaluating the specified object by minimizing each evaluation dimension and the label corresponding to the comment data as an optimization target.
Optionally, the pre-constructing a knowledge graph corresponding to each evaluation dimension specifically includes:
acquiring each evaluation dimension corresponding to an application scene;
determining a central node corresponding to each evaluation dimension according to the word corresponding to each evaluation dimension;
for each evaluation dimension, inputting the comment data and the central node corresponding to the evaluation dimension into a pre-trained scoring model, and determining a first-order neighbor node of the central node corresponding to the evaluation dimension and a second-order neighbor node of the central node corresponding to the evaluation dimension;
and constructing a knowledge graph corresponding to the evaluation dimension according to the central node corresponding to the evaluation dimension, the first-order neighbor node of the central node corresponding to the evaluation dimension and the second-order neighbor node of the central node corresponding to the evaluation dimension.
Optionally, for each evaluation dimension, inputting the comment data and the center node corresponding to the evaluation dimension into a pre-trained scoring model, and determining a first-order neighbor node of the center node corresponding to the evaluation dimension and a second-order neighbor node of the center node corresponding to the evaluation dimension, specifically including:
acquiring a general knowledge graph;
for each evaluation dimension, inputting the comment data and the central node corresponding to the evaluation dimension into a pre-trained scoring model, determining a relevance score corresponding to each word in the comment data according to the relevance between the word corresponding to the evaluation dimension and each word in the comment data, and taking the word of which the relevance score is greater than a set scoring threshold value as a first-order neighbor node of the central node corresponding to the evaluation dimension;
and inquiring the first-order neighbor node in the general knowledge graph aiming at each first-order neighbor node, determining words with the occurrence frequency within a set frequency range in the inquiring process, and taking the words as second-order neighbor nodes of the central node corresponding to the evaluation dimension.
Optionally, the dimensional characteristic data comprises overall dimensional characteristic data;
for each evaluation dimension, inputting the comment data and the knowledge graph corresponding to the evaluation dimension into a prediction model to be trained to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension, specifically including:
for each evaluation dimension, inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a prediction model to be trained so as to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension;
and splicing the feature data of each node in the knowledge graph corresponding to the evaluation dimension, and determining the overall dimension feature data corresponding to the evaluation dimension.
Optionally, the dimension feature data includes central dimension feature data, the prediction model includes a multi-channel graph convolution network, and different channels correspond to knowledge maps corresponding to different evaluation dimensions;
for each evaluation dimension, inputting the comment data and the knowledge graph corresponding to the evaluation dimension into a prediction model to be trained to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension, specifically including:
for each evaluation dimension, inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a prediction model to be trained so as to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension;
inputting the feature data of each node in the knowledge graph corresponding to the evaluation dimension into a channel corresponding to the evaluation dimension in the multi-channel graph convolution network, and determining the central dimension feature data corresponding to the evaluation dimension according to the incidence relation among the nodes in the knowledge graph and the feature data corresponding to the nodes.
Optionally, the prediction model includes multiple association weight layers, and different evaluation dimensions correspond to the different association weight layers;
determining an evaluation result of the user evaluating the specified object by the evaluation dimension according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data, specifically comprising:
inputting the overall dimension characteristic data corresponding to the evaluation dimension and the evaluation characteristic data into the association weight layer corresponding to the evaluation dimension, and determining the association characteristic data corresponding to the evaluation dimension;
and determining an evaluation result of the user for evaluating the specified object by the evaluation dimension according to the associated feature data corresponding to the evaluation dimension and the central dimension feature data corresponding to the evaluation dimension.
Optionally, determining, according to the associated feature data corresponding to the evaluation dimension and the center dimension feature data corresponding to the evaluation dimension, an evaluation result of the user evaluating the specified object with the evaluation dimension, specifically including:
acquiring a feature weight corresponding to each evaluation dimension, wherein the feature weight is used for balancing the influence of the associated feature data corresponding to each evaluation dimension and the central dimension feature data corresponding to each evaluation dimension on an evaluation result;
for each evaluation dimension, determining fusion feature data corresponding to the evaluation dimension according to the associated feature data corresponding to the evaluation dimension, the center dimension feature data corresponding to the evaluation dimension and the feature weight corresponding to the evaluation dimension;
and according to the fusion characteristic data corresponding to the evaluation dimension, the user evaluates the specified object according to the evaluation dimension.
The present specification provides a method for information recommendation, including:
obtaining comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension;
inputting the comment data and the knowledge graph corresponding to the evaluation dimension into a trained prediction model aiming at each evaluation dimension to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension, wherein the prediction model is obtained by training through the model training method;
determining an evaluation result of the user for evaluating the specified object by the evaluation dimension according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data;
and recommending information to the user according to the evaluation result of evaluating the specified object according to each evaluation dimension.
The present specification provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the above-described method of model training and method of information recommendation.
The present specification provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method for model training and the method for information recommendation when executing the program.
The technical scheme adopted by the specification can achieve the following beneficial effects:
in the model training method provided by the specification, a server can acquire comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension. Secondly, inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a prediction model to be trained aiming at each evaluation dimension so as to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension. And then, according to the dimension characteristic data and the evaluation characteristic data corresponding to the evaluation dimension, determining an evaluation result of the user for evaluating the specified object by the evaluation dimension. And finally, training the prediction model by taking the deviation between the evaluation result of evaluating the specified object by minimizing each evaluation dimension and the label corresponding to the comment data as an optimization target.
According to the method, the comment data and the corresponding knowledge graph under each evaluation dimension can be input into the prediction model, the evaluation result of the user for evaluating the specified object by each evaluation dimension is determined, and the prediction model is trained by taking the minimized deviation between the evaluation result of the specified object evaluated by each evaluation dimension and the label corresponding to the comment data as an optimization target. Therefore, the prediction model can determine accurate evaluation results corresponding to each evaluation dimension, information recommendation is carried out on the user according to the evaluation results corresponding to each evaluation dimension, and the purchase experience of the user is effectively improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the specification and are incorporated in and constitute a part of this specification, illustrate embodiments of the specification and together with the description serve to explain the specification and not to limit the specification in a non-limiting sense. In the drawings:
FIG. 1 is a schematic flow chart of a method of model training in the present specification;
FIG. 2 is a schematic diagram of a model structure of a prediction model in the present specification;
FIG. 3 is a schematic diagram of a method for information recommendation in the present specification;
FIG. 4 is a schematic diagram of an apparatus for model training in the present specification;
FIG. 5 is a schematic diagram of an information recommendation device in the present specification;
fig. 6 is a schematic diagram of an electronic device corresponding to fig. 1 provided in the present specification.
Detailed Description
In order to make the objects, technical solutions and advantages of the present disclosure more clear, the technical solutions of the present disclosure will be clearly and completely described below with reference to the specific embodiments of the present disclosure and the accompanying drawings. It is to be understood that the embodiments described are only a few embodiments of the present disclosure, and not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present specification without making any creative effort belong to the protection scope of the present specification.
The technical solutions provided by the embodiments of the present description are described in detail below with reference to the accompanying drawings.
Fig. 1 is a schematic flow chart of a model training method in this specification, which specifically includes the following steps:
s100: and obtaining comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension.
The execution subject of the model training provided in the present specification may be a terminal device such as a server, a device cluster composed of a plurality of computers, or the like. For convenience of description, the method for model training provided in this specification will be described below with only a server as an execution subject.
In this specification, a server may acquire comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension. The designated object referred to herein may refer to a product of a set category. For example, a category is set to restaurant, and the designated object is a restaurant of a different brand. Reference herein to an evaluation dimension may refer to evaluating different aspects of a specified object. For example, specifying an object as a brand a restaurant, the evaluation dimensions include: service, environment, taste, etc.
In practical application, a prediction model is usually trained through comment data of a user in a text form, the method has a poor training effect on the prediction model, and the accuracy of the determined evaluation result of the user in each evaluation dimension is low. Based on the method, the server can construct the knowledge graph corresponding to each evaluation dimension to be used for model training of the prediction model, and the training effect of the prediction model is improved.
In this embodiment, the server may obtain each evaluation dimension corresponding to the application scenario. The application scenes mentioned here can refer to scenes of restaurants, hotels, tourist attractions, etc.
Secondly, the server can determine the central node corresponding to each evaluation dimension according to the word corresponding to each evaluation dimension. For example, the application scenario is a restaurant, one evaluation dimension is a service, and a word corresponding to the evaluation dimension is a service.
Then, for each evaluation dimension, the server may input the comment data and the center node corresponding to the evaluation dimension into a pre-trained scoring model, and determine a first-order neighbor node of the center node corresponding to the evaluation dimension and a second-order neighbor node of the center node corresponding to the evaluation dimension.
Finally, the server can construct the knowledge graph corresponding to the evaluation dimension according to the central node corresponding to the evaluation dimension, the first-order neighbor node of the central node corresponding to the evaluation dimension and the second-order neighbor node of the central node corresponding to the evaluation dimension.
In practical application, if the knowledge graph corresponding to each evaluation dimension is constructed only according to the comment data, common sense semantic information covered by the user in daily expression cannot be obtained. Therefore, the server can acquire the universal knowledge graph, and the universal knowledge graph is combined on the basis of the knowledge graph constructed by the comment data so as to obtain the common sense semantic information covered by the user in daily expression.
In embodiments of the present description, the server may obtain a generic knowledge graph. The general Knowledge Graph referred to herein may refer to at least one of an Encyclopedia knowledgebase Graph (Encyclopedia knowledgebase Graph) and a Common Sense knowledgebase Graph (Common Sense knowledgebase Graph).
Secondly, the server inputs the comment data and the center node corresponding to the evaluation dimension into a pre-trained scoring model according to each evaluation dimension, determines a relevance score corresponding to each word in the comment data according to the relevance between the word corresponding to the evaluation dimension and each word in the comment data, and takes the word with the relevance score larger than a set scoring threshold value as a first-order neighbor node of the center node corresponding to the evaluation dimension.
For example, the central node corresponding to the comment dimension is a service, the comment data includes phrases such as good service, weekly arrival of service, and popular service, the scoring model may determine relevance scores of words such as good service, weekly arrival, and popular service according to relevance between words such as good service, weekly arrival, and popular service and the service, and if a word whose relevance score of words such as good service, weekly arrival, and popular service is greater than a set scoring threshold is determined, the word such as good service, weekly arrival, and popular service is used as a first-order neighbor node of the central node corresponding to the comment dimension.
In practical application, the general knowledge graph is composed of massive triples, and when a first-order neighbor node is queried, if the frequency of the related partial words is less in the process of querying the first-order neighbor node, the words are rarely used. If the number of times of the related partial words is excessive in the process of inquiring the first-order neighbor node, the words are common, and the training effect in the subsequent model training process is poor. Therefore, the server can construct the knowledge graph corresponding to each evaluation dimension through the words with the occurrence frequency within the set frequency range.
The server can inquire the first-order neighbor node in the general knowledge graph aiming at each first-order neighbor node, determine words with the occurrence frequency within a set frequency range in the inquiry process, and take the words as second-order neighbor nodes of the central node corresponding to the evaluation dimension.
S102: and inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a to-be-trained prediction model aiming at each evaluation dimension so as to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension.
In this specification, the server may input, for each evaluation dimension, the comment data and the corresponding knowledge graph in the evaluation dimension into the prediction model to be trained, so as to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension.
In this specification, the dimension feature data includes overall dimension feature data, and the server may input, for each evaluation dimension, the comment data and the corresponding knowledge graph in the evaluation dimension into the prediction model to be trained, so as to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension.
Secondly, the server can splice the feature data of each node in the knowledge graph corresponding to the evaluation dimension to determine the overall dimension feature data corresponding to the evaluation dimension.
In the embodiment of the present specification, the dimension feature data includes center dimension feature data, the prediction model includes a multi-channel graph convolution network, and different channels correspond to knowledge maps corresponding to different evaluation dimensions. The server can input the comment data and the corresponding knowledge graph under the evaluation dimension into the prediction model to be trained aiming at each evaluation dimension so as to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension.
Secondly, the server can input the feature data of each node in the knowledge graph corresponding to the evaluation dimension into a channel corresponding to the evaluation dimension in the multi-channel graph convolution network, and the center dimension feature data corresponding to the evaluation dimension is determined according to the incidence relation among the nodes in the knowledge graph and the feature data corresponding to the nodes.
It should be noted that the central dimension feature data corresponding to the evaluation dimension is used for characterizing feature data corresponding to the evaluation dimension, which is obtained by combining common sense semantic information covered by the user in daily expression associated with the word corresponding to the evaluation dimension.
S104: and determining an evaluation result of the user evaluating the specified object by the evaluation dimension according to the dimension characteristic data corresponding to the evaluation dimension and the evaluation characteristic data.
In this embodiment, the server may determine, according to the dimension feature data and the evaluation feature data corresponding to the evaluation dimension, an evaluation result of the user evaluating the specified object in the evaluation dimension.
In practical applications, because different users have different language habits, phrases with similar meanings may appear in comment data of different users. For example, phrases that work well, do things well, serve well, etc., all mean that the service is good. However, the current prediction model only judges the phrases to be good comments, and cannot judge which comment dimension the phrases belong to. Based on the above, the server may set an association weight layer corresponding to each evaluation dimension, and for each evaluation dimension, determine, from the overall dimension feature data and the comment feature data corresponding to the evaluation dimension, feature data associated with the evaluation dimension through the association weight layer corresponding to the evaluation dimension.
In the embodiment of the present specification, the prediction model includes a plurality of association weight layers, and different association weight layers correspond to different evaluation dimensions. The server may input the overall dimension characteristic data and the evaluation characteristic data corresponding to the evaluation dimension into the association weight layer corresponding to the evaluation dimension, and determine the association characteristic data corresponding to the evaluation dimension.
Then, the server may determine, according to the associated feature data corresponding to the evaluation dimension and the center dimension feature data corresponding to the evaluation dimension, an evaluation result that the user evaluates the specified object with the evaluation dimension.
In practical applications, for each evaluation dimension, the influence of the associated feature data corresponding to the evaluation dimension and the influence of the central dimension feature data corresponding to the evaluation dimension on the evaluation result are different, and therefore, the server needs to balance the influences of the associated feature data and the central dimension feature data on the evaluation result.
In this embodiment, the server may obtain a feature weight corresponding to each evaluation dimension, where the feature weight is used to balance an influence of associated feature data corresponding to each evaluation dimension and center dimension feature data corresponding to each evaluation dimension on an evaluation result.
Secondly, for each evaluation dimension, the server may determine fusion feature data corresponding to the evaluation dimension according to the associated feature data corresponding to the evaluation dimension, the center dimension feature data corresponding to the evaluation dimension, and the feature weight corresponding to the evaluation dimension.
Finally, the server can determine the evaluation result of the user evaluating the specified object by the evaluation dimension according to the fusion characteristic data corresponding to the evaluation dimension.
The method for converting the candidate recommendation information into the feature data may be various, for example, a transformer-based Bidirectional Encoder Representation from transforms (BERT) model, a Bidirectional Long Short-Term Memory (Bi-LSTM) model, a brute force Optimized BERT model (RoBERTa), and the like, and this specification is not limited herein. If the method used is a brute force optimization BERT model, the feature data of each word in the comment data and the feature data of the position of each word in the comment data are determined, and the feature data corresponding to the comment data are finally determined according to the feature data. And determining feature data of each node in the knowledge-graph.
The method for extracting feature data from the knowledge Graph corresponding to each evaluation dimension may be a Graph Convolutional Network (GCN), a Multi-channel Graph Convolutional Network (MGCN), or the like, and this specification is not limited herein. For each evaluation dimension, the feature data of each node in the knowledge graph corresponding to the evaluation dimension is spliced, and the feature data can be realized by adopting a concat layer of a basic model, a weighted-sum layer of a Deep Interest Network (Deep Interest Network, DIN) model (the feature data of each node is spliced through a weight parameter), and other Network structures, and the description is not limited herein.
S106: and training the prediction model by taking the deviation between the evaluation result of evaluating the specified object by minimizing each evaluation dimension and the label corresponding to the comment data as an optimization target.
In the embodiment of the present specification, the server may train the prediction model with a goal of minimizing a deviation between an evaluation result obtained by evaluating the specified object by each evaluation dimension and a label corresponding to the comment data as an optimization goal. The label corresponding to the comment data mentioned herein may refer to an actual result of the comment data in each evaluation dimension. For example, specifying an object as a brand a restaurant, the evaluation dimensions include: service, environment, taste and the like, and the labels corresponding to the comment data are service favorable comment, environment favorable comment and taste bad comment.
In this embodiment, the server may determine the loss and the value according to the deviation between the evaluation result of evaluating the specified object by each evaluation dimension and the label corresponding to the comment data, and train the prediction model with the minimized loss and the minimized value as the optimization target.
In the embodiment of the present specification, the model structure of the prediction model includes: the specific process of the multi-channel graph convolution network, the characteristic weight layer and the association weight layer is shown in fig. 2.
Fig. 2 is a schematic diagram of a model structure of a prediction model in this specification.
In fig. 2, the server constructs a knowledge graph corresponding to each evaluation dimension by using the general knowledge graph and each word corresponding to each evaluation dimension. And inputting the knowledge graph corresponding to each evaluation dimension into the feature layer, and determining feature data of each node in the knowledge graph corresponding to each evaluation dimension. And splicing the feature data of each node in the knowledge graph corresponding to each evaluation dimension to obtain the overall dimension feature data corresponding to the evaluation dimension. And inputting the feature data of each node in the knowledge graph corresponding to the evaluation dimension into the graph convolution network corresponding to the evaluation dimension to obtain the central dimension feature data corresponding to the evaluation dimension.
Secondly, the server can input the overall dimension feature data and the comment feature data corresponding to the evaluation dimension into the association weight layer corresponding to the evaluation dimension to obtain the association feature data corresponding to the evaluation dimension.
Then, the server may input the associated feature data corresponding to the evaluation dimension and the center dimension feature data corresponding to the evaluation dimension to the feature weight layer corresponding to the evaluation dimension to obtain the fused feature data corresponding to the evaluation dimension.
Finally, the server can determine the evaluation result corresponding to the evaluation dimension according to the fusion feature data corresponding to the evaluation dimension.
It can be seen from the above contents that the method can input the comment data and the corresponding knowledge graph under each evaluation dimension into the prediction model, determine the evaluation result of the user evaluating the specified object with each evaluation dimension, and train the prediction model with minimizing the deviation between the evaluation result of the specified object evaluated with each evaluation dimension and the label corresponding to the comment data as the optimization target. Therefore, the prediction model can determine accurate evaluation results corresponding to each evaluation dimension, information recommendation is carried out on the user according to the evaluation results corresponding to the evaluation dimensions, and the purchase experience of the user is effectively improved.
After the training of the prediction model is completed, the embodiment of the description can recommend information to the user through the prediction model, and a specific process is shown in fig. 3.
Fig. 3 is a flowchart illustrating a method for information recommendation in this specification.
S300: and obtaining comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension.
S302: and inputting the comment data and the knowledge graph corresponding to the evaluation dimension into a trained prediction model aiming at each evaluation dimension to determine evaluation characteristic data corresponding to the comment data and dimension characteristic data corresponding to the evaluation dimension, wherein the prediction model is obtained by training through the model training method.
S304: and determining an evaluation result of the user evaluating the specified object by the evaluation dimension according to the dimension characteristic data corresponding to the evaluation dimension and the evaluation characteristic data.
S306: and recommending information to the user according to the evaluation result of evaluating the specified object according to each evaluation dimension.
In this specification embodiment, a server may acquire comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension.
Secondly, for each evaluation dimension, the server may input the comment data and the knowledge graph corresponding to the evaluation dimension into the trained prediction model to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension.
Then, the server can determine an evaluation result of the user evaluating the specified object with the evaluation dimension according to the dimension feature data and the evaluation feature data corresponding to the evaluation dimension.
Finally, the server can recommend information to the user according to the evaluation result of the evaluation of each evaluation dimension on the designated object.
In this embodiment of the present specification, if the specified object is an a restaurant, the server may acquire each comment data for the a restaurant, input each comment data and a knowledge graph of each evaluation dimension corresponding to the restaurant into the trained prediction model, and determine an evaluation result of each evaluation dimension evaluating the a restaurant.
And determining the evaluation result of which the occurrence frequency is greater than a set frequency threshold value under each evaluation dimension as a final evaluation result corresponding to the evaluation dimension. And displaying the final evaluation result corresponding to each evaluation dimension to the user. For example, if the evaluation dimension is a service, and the evaluation result with the largest occurrence number in the evaluation dimension is a good service, the final evaluation result corresponding to the service is a good service. The evaluation dimension is a cost performance, the evaluation result with the largest occurrence frequency under the evaluation dimension is a high cost performance, and the final evaluation result corresponding to the cost performance is the high cost performance. Finally, the service is displayed to the user with good service and high cost performance.
Based on the same idea, the present specification further provides a corresponding model training apparatus, as shown in fig. 4.
Fig. 4 is a schematic diagram of a model training apparatus provided in this specification, which specifically includes:
the obtaining module 400 is configured to obtain comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension;
an input module 402, configured to input, for each evaluation dimension, the comment data and the knowledge graph corresponding to the evaluation dimension into a prediction model to be trained, so as to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension;
a determining module 404, configured to determine, according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data, an evaluation result that the user evaluates the specified object with the evaluation dimension;
a training module 406, configured to train the prediction model by taking a deviation between an evaluation result obtained by evaluating the specified object with minimized evaluation dimensions and a label corresponding to the comment data as an optimization target.
Optionally, the obtaining module 400 is specifically configured to obtain each evaluation dimension corresponding to an application scenario, determine a central node corresponding to each evaluation dimension according to a word corresponding to each evaluation dimension, input the comment data and the central node corresponding to the evaluation dimension to a pre-trained scoring model for each evaluation dimension, determine a first-order neighbor node of the central node corresponding to the evaluation dimension and a second-order neighbor node of the central node corresponding to the evaluation dimension, and construct a knowledge graph corresponding to the evaluation dimension according to the central node corresponding to the evaluation dimension, the first-order neighbor node of the central node corresponding to the evaluation dimension, and the second-order neighbor node of the central node corresponding to the evaluation dimension.
Optionally, the obtaining module 400 is specifically configured to obtain a general knowledge graph, input the comment data and the center node corresponding to the evaluation dimension into a pre-trained score model for each evaluation dimension, determine a relevance score corresponding to each word in the comment data according to a relevance between the word corresponding to the evaluation dimension and each word in the comment data, regard a word whose relevance score is greater than a set score threshold as a first-order neighbor node of the center node corresponding to the evaluation dimension, query, for each first-order neighbor node, the first-order neighbor node in the general knowledge graph, and determine a word whose occurrence frequency is within a set frequency range in a query process, as a second-order neighbor node of the center node corresponding to the evaluation dimension.
Optionally, the dimensional characteristic data comprises overall dimensional characteristic data;
the input module 402 is specifically configured to, for each evaluation dimension, input the comment data and the corresponding knowledge graph in the evaluation dimension into a prediction model to be trained to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension, splice the feature data of each node in the knowledge graph corresponding to the evaluation dimension, and determine overall dimension feature data corresponding to the evaluation dimension.
Optionally, the dimension feature data includes central dimension feature data, the prediction model includes a multi-channel graph convolution network, and different channels correspond to knowledge maps corresponding to different evaluation dimensions;
the input module 402 is specifically configured to, for each evaluation dimension, input the comment data and the corresponding knowledge graph in the evaluation dimension into a prediction model to be trained to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension, input the feature data of each node in the knowledge graph corresponding to the evaluation dimension into a channel corresponding to the evaluation dimension in the multi-channel graph convolution network, and determine central dimension feature data corresponding to the evaluation dimension according to an association relationship between each node in the knowledge graph and the feature data corresponding to each node.
Optionally, the prediction model includes a plurality of association weight layers, and different evaluation dimensions correspond to different association weight layers;
the determining module 404 is specifically configured to input the overall dimension feature data corresponding to the evaluation dimension and the evaluation feature data into the association weight layer corresponding to the evaluation dimension, determine the association feature data corresponding to the evaluation dimension, and determine an evaluation result of the user evaluating the designated object with the evaluation dimension according to the association feature data corresponding to the evaluation dimension and the center dimension feature data corresponding to the evaluation dimension.
Optionally, the determining module 404 is specifically configured to obtain a feature weight corresponding to each evaluation dimension, where the feature weight is used to balance an influence of associated feature data corresponding to each evaluation dimension and center dimension feature data corresponding to each evaluation dimension on an evaluation result, determine, for each evaluation dimension, fused feature data corresponding to the evaluation dimension according to the associated feature data corresponding to the evaluation dimension, the center dimension feature data corresponding to the evaluation dimension, and the feature weight corresponding to the evaluation dimension, and determine, according to the fused feature data corresponding to the evaluation dimension, an evaluation result of the user for evaluating the designated object with the evaluation dimension.
Based on the same idea, the present specification further provides a corresponding information recommendation apparatus, as shown in fig. 5.
Fig. 5 is a schematic diagram of an information recommendation apparatus provided in this specification, which specifically includes:
the acquisition module 500 is configured to acquire comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension;
an input module 502, configured to input, for each evaluation dimension, the comment data and the knowledge graph corresponding to the evaluation dimension into a trained prediction model to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension, where the prediction model is obtained by training through a model training method;
a determining module 504, configured to determine, according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data, an evaluation result that the user evaluates the specified object with the evaluation dimension
And a recommending module 506, configured to recommend information to the user according to an evaluation result obtained by evaluating the specified object according to each evaluation dimension.
The present specification also provides a computer-readable storage medium storing a computer program, which can be used to execute the method for model training and the method for information recommendation shown in fig. 1.
This specification also provides a schematic block diagram of the electronic device shown in fig. 6. As shown in fig. 6, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include hardware required for other services. The processor reads a corresponding computer program from the non-volatile memory into the memory and then runs the computer program to implement the model training method described in fig. 1 and the information recommendation method described in fig. 3. Of course, besides the software implementation, the present specification does not exclude other implementations, such as logic devices or a combination of software and hardware, and the like, that is, the execution subject of the following processing flow is not limited to each logic unit, and may be hardware or logic devices.
In the 90 s of the 20 th century, improvements in a technology could clearly distinguish between improvements in hardware (e.g., improvements in circuit structures such as diodes, transistors, switches, etc.) and improvements in software (improvements in process flow). However, as technology advances, many of today's process flow improvements have been seen as direct improvements in hardware circuit architecture. Designers almost always obtain a corresponding hardware circuit structure by programming an improved method flow into the hardware circuit. Thus, it cannot be said that an improvement in the process flow cannot be realized by hardware physical blocks. For example, a Programmable Logic Device (PLD), such as a Field Programmable Gate Array (FPGA), is an integrated circuit whose Logic functions are determined by programming the Device by a user. A digital system is "integrated" on a PLD by the designer's own programming without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing an Integrated Circuit chip, such Programming is often implemented by "logic compiler" software, which is similar to a software compiler used in program development and writing, but the original code before compiling is also written by a specific Programming Language, which is called Hardware Description Language (HDL), and HDL is not only one but many, such as ABEL (Advanced Boolean Expression Language), AHDL (alternate Hardware Description Language), traffic, CUPL (core universal Programming Language), HDCal, jhddl (Java Hardware Description Language), lava, lola, HDL, PALASM, rhyd (Hardware Description Language), and vhigh-Language (Hardware Description Language), which is currently used in most popular applications. It will also be apparent to those skilled in the art that hardware circuitry that implements the logical method flows can be readily obtained by merely slightly programming the method flows into an integrated circuit using the hardware description languages described above.
The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, atmel AT91SAM, microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic for the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may thus be considered a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.
The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
For convenience of description, the above devices are described as being divided into various units by function, respectively. Of course, the functionality of the various elements may be implemented in the same one or more pieces of software and/or hardware in the practice of this description.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both permanent and non-permanent, removable and non-removable media, may implement the information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of other like elements in a process, method, article, or apparatus comprising the element.
As will be appreciated by one skilled in the art, embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the description may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the description may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
This description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The above description is only an example of the present disclosure, and is not intended to limit the present disclosure. Various modifications and alterations to this description will become apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification should be included in the scope of the claims of the present specification.
Claims (10)
1. A method of model training, comprising:
obtaining comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension;
for each evaluation dimension, inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a prediction model to be trained so as to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension;
determining an evaluation result of the user for evaluating the specified object by the evaluation dimension according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data;
and training the prediction model by taking the deviation between the evaluation result of evaluating the specified object by minimizing each evaluation dimension and the label corresponding to the comment data as an optimization target.
2. The method according to claim 1, wherein the pre-constructing of the knowledge graph corresponding to each evaluation dimension specifically comprises:
acquiring each evaluation dimension corresponding to an application scene;
determining a central node corresponding to each evaluation dimension according to the word corresponding to each evaluation dimension;
for each evaluation dimension, inputting the comment data and the central node corresponding to the evaluation dimension into a pre-trained scoring model, and determining a first-order neighbor node of the central node corresponding to the evaluation dimension and a second-order neighbor node of the central node corresponding to the evaluation dimension;
and constructing a knowledge graph corresponding to the evaluation dimension according to the central node corresponding to the evaluation dimension, the first-order neighbor node of the central node corresponding to the evaluation dimension and the second-order neighbor node of the central node corresponding to the evaluation dimension.
3. The method of claim 2, wherein for each evaluation dimension, inputting the comment data and the central node corresponding to the evaluation dimension into a pre-trained scoring model, and determining a first-order neighbor node of the central node corresponding to the evaluation dimension and a second-order neighbor node of the central node corresponding to the evaluation dimension specifically includes:
acquiring a general knowledge graph;
for each evaluation dimension, inputting the comment data and the central node corresponding to the evaluation dimension into a pre-trained scoring model, determining a relevance score corresponding to each word in the comment data according to the relevance between the word corresponding to the evaluation dimension and each word in the comment data, and taking the word of which the relevance score is greater than a set scoring threshold value as a first-order neighbor node of the central node corresponding to the evaluation dimension;
and inquiring the first-order neighbor node in the general knowledge graph aiming at each first-order neighbor node, determining words with the occurrence frequency within a set frequency range in the inquiring process, and taking the words as second-order neighbor nodes of the central node corresponding to the evaluation dimension.
4. The method of claim 1, wherein the dimensional feature data comprises overall dimensional feature data;
for each evaluation dimension, inputting the comment data and the knowledge graph corresponding to the evaluation dimension into a prediction model to be trained to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension, specifically including:
for each evaluation dimension, inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a prediction model to be trained so as to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension;
and splicing the feature data of each node in the knowledge graph corresponding to the evaluation dimension, and determining the overall dimension feature data corresponding to the evaluation dimension.
5. The method of claim 4, wherein the dimensional feature data comprises central dimensional feature data, the prediction model comprises a multi-channel graph convolution network, and different channels correspond to knowledge graphs corresponding to different evaluation dimensions;
for each evaluation dimension, inputting the comment data and the knowledge graph corresponding to the evaluation dimension into a prediction model to be trained to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension, specifically including:
for each evaluation dimension, inputting the comment data and the corresponding knowledge graph under the evaluation dimension into a prediction model to be trained so as to determine evaluation feature data corresponding to the comment data and feature data of each node in the knowledge graph corresponding to the evaluation dimension;
inputting the feature data of each node in the knowledge graph corresponding to the evaluation dimension into a channel corresponding to the evaluation dimension in the multi-channel graph convolution network, and determining the central dimension feature data corresponding to the evaluation dimension according to the association relation between the nodes in the knowledge graph and the feature data corresponding to the nodes.
6. The method of claim 5, wherein the predictive model includes a plurality of layers of correlation weights, wherein different evaluation dimensions correspond to different layers of correlation weights;
determining an evaluation result of the user evaluating the specified object by the evaluation dimension according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data, specifically comprising:
inputting the overall dimension characteristic data corresponding to the evaluation dimension and the evaluation characteristic data into the association weight layer corresponding to the evaluation dimension, and determining the association characteristic data corresponding to the evaluation dimension;
and determining an evaluation result of the user for evaluating the specified object by the evaluation dimension according to the associated feature data corresponding to the evaluation dimension and the central dimension feature data corresponding to the evaluation dimension.
7. The method according to claim 6, wherein determining an evaluation result of the user evaluating the specified object with the evaluation dimension according to the associated feature data corresponding to the evaluation dimension and the center dimension feature data corresponding to the evaluation dimension specifically comprises:
acquiring a feature weight corresponding to each evaluation dimension, wherein the feature weight is used for balancing the influence of the associated feature data corresponding to each evaluation dimension and the central dimension feature data corresponding to each evaluation dimension on an evaluation result;
for each evaluation dimension, determining fusion feature data corresponding to the evaluation dimension according to the associated feature data corresponding to the evaluation dimension, the center dimension feature data corresponding to the evaluation dimension and the feature weight corresponding to the evaluation dimension;
and determining an evaluation result of the user evaluating the specified object according to the fusion feature data corresponding to the evaluation dimension.
8. A method for information recommendation, comprising:
obtaining comment data of a user for a specified object and a pre-constructed knowledge graph corresponding to each evaluation dimension;
inputting the comment data and the knowledge graph corresponding to the evaluation dimension into a trained prediction model for each evaluation dimension to determine evaluation feature data corresponding to the comment data and dimension feature data corresponding to the evaluation dimension, wherein the prediction model is obtained by training through the method of any one of claims 1 to 7;
determining an evaluation result of the user for evaluating the specified object by the evaluation dimension according to the dimension feature data corresponding to the evaluation dimension and the evaluation feature data;
and recommending information to the user according to the evaluation result of evaluating the specified object according to each evaluation dimension.
9. A computer-readable storage medium, characterized in that the storage medium stores a computer program which, when executed by a processor, implements the method of any of the preceding claims 1 to 8.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any of claims 1 to 8 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210851513.8A CN115168726A (en) | 2022-07-19 | 2022-07-19 | Model training method and information recommendation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210851513.8A CN115168726A (en) | 2022-07-19 | 2022-07-19 | Model training method and information recommendation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115168726A true CN115168726A (en) | 2022-10-11 |
Family
ID=83494332
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210851513.8A Pending CN115168726A (en) | 2022-07-19 | 2022-07-19 | Model training method and information recommendation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115168726A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116402113A (en) * | 2023-06-08 | 2023-07-07 | 之江实验室 | Task execution method and device, storage medium and electronic equipment |
-
2022
- 2022-07-19 CN CN202210851513.8A patent/CN115168726A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116402113A (en) * | 2023-06-08 | 2023-07-07 | 之江实验室 | Task execution method and device, storage medium and electronic equipment |
| CN116402113B (en) * | 2023-06-08 | 2023-10-03 | 之江实验室 | Task execution method and device, storage medium and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110309283A (en) | A kind of answer of intelligent answer determines method and device | |
| KR20180107136A (en) | Method and system for search engine selection and optimization | |
| CN113688313A (en) | Training method of prediction model, information pushing method and device | |
| CN112733024A (en) | Information recommendation method and device | |
| CN115203394A (en) | Model training method, service execution method and device | |
| CN113344098A (en) | Model training method and device | |
| CN110532295A (en) | A kind of method and device of computer-implemented information processing, information inquiry | |
| CN115187311A (en) | Shop site selection method and device suitable for multiple industries | |
| CN115168726A (en) | Model training method and information recommendation method | |
| CN115455166A (en) | Method, device, medium and equipment for detecting abnormality of intelligent dialogue system | |
| CN113887206B (en) | Model training and keyword extraction method and device | |
| CN112559575A (en) | Search processing method, customer service information processing method and device | |
| CN113641894A (en) | Information recommendation method and device | |
| CN112948449A (en) | Information recommendation method and device | |
| CN113343095A (en) | Model training and information recommendation method and device | |
| CN113343085B (en) | Information recommendation method and device, storage medium and electronic equipment | |
| CN113204637A (en) | Text processing method and device, storage medium and electronic equipment | |
| CN112699307A (en) | Information generation method and device | |
| CN116701570A (en) | Recall result screening method, recall result screening device, computer equipment and storage medium | |
| CN113887234B (en) | Model training and recommending method and device | |
| CN114116816A (en) | Recommendation method and device | |
| CN113343132B (en) | Model training method, information display method and device | |
| CN115689676A (en) | Industry identification method and apparatus, storage medium and electronic device | |
| CN114996570A (en) | Information recommendation method and device | |
| CN114331602A (en) | Model training method based on transfer learning, information recommendation method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |