CN110968698B - User vehicle using experience investigation method and system based on knowledge graph and cloud server - Google Patents
User vehicle using experience investigation method and system based on knowledge graph and cloud server Download PDFInfo
- Publication number
- CN110968698B CN110968698B CN201811146863.4A CN201811146863A CN110968698B CN 110968698 B CN110968698 B CN 110968698B CN 201811146863 A CN201811146863 A CN 201811146863A CN 110968698 B CN110968698 B CN 110968698B
- Authority
- CN
- China
- Prior art keywords
- user
- experience
- intention
- structure model
- network
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
Abstract
The application provides a user vehicle using experience investigation method, a system and a cloud server based on a knowledge graph. According to the method and the device, the judgment model can be built by utilizing the algorithm network, and the parameters of the judgment model can be continuously corrected according to the actual situation, so that the score can be automatically obtained, the effective information of the real intention of the user can be intelligently identified according to the score, the development requirement can be made by research personnel according to the effective information, and the product can adapt to the development trend of market requirements.
Description
Technical Field
The application relates to the technical field of data research, in particular to a user vehicle using experience research method based on a knowledge graph, a user vehicle using experience research system based on the knowledge graph, and a cloud server for configuring the user vehicle using experience research system based on the knowledge graph.
Background
With the rapid development of the internet, people gradually enter an information overload era from an information shortage era. The problem of information flooding in the network becomes extremely serious due to the explosive growth of information, so that it is difficult for users to find valuable data from massive data, and some useful information is often submerged in the sea of information to become island information.
In the field of automotive technology, research and development personnel often need to investigate the satisfaction degree of users when developing, and perform targeted design according to the satisfaction degree of users. However, many users are not really intended by the user, such as many users do not want to spend time, and therefore draw a tick or write some comments randomly on the survey form.
These are not real user experiences, which not only do nothing to the developers, but also adversely affect the research and development judgment, resulting in the production of products which are completely contrary to the market demand, and seriously affect the subsequent product sales, even the life and death of the company.
Aiming at the defects in various aspects of the prior art, the inventor of the application provides a user vehicle using experience investigation method and system based on a knowledge graph and a cloud server through deep research.
Disclosure of Invention
The application aims to provide a user vehicle using experience investigation method, a user vehicle using experience investigation system and a cloud server based on a knowledge graph, a judgment model can be built by utilizing an algorithm network, parameters of the judgment model can be continuously corrected according to actual conditions, so that the judgment model can be used for automatically obtaining scores, effective information of real intentions of users can be intelligently identified according to the scores, development requirements can be made by research and development personnel according to the effective information, the product can adapt to the development trend of market demands, and the product competitiveness of companies is improved.
In order to solve the technical problem, the present application provides a knowledge graph-based user vehicle usage experience investigation method, as one implementation manner, the user vehicle usage experience investigation method includes the steps of:
establishing a network prediction structure model;
inputting scoring information used for representing vehicle experience of a user into the network prediction structure model, and judging fuzzy intention of the scoring information through the network prediction structure model;
Inputting the scoring information into a knowledge graph database, and judging a standard intention of the scoring information by using the knowledge graph database;
comparing the fuzzy intention with the standard intention, and establishing a loss function of the scoring information according to the standard intention;
and adjusting the structural parameters of the network prediction structure model according to the loss function, and updating the adjusted structural parameters to the network prediction structure model to obtain an expected classifier function for judging the actual user vehicle using experience intention.
As one embodiment, before the step of inputting the scoring information representing the user experience with the vehicle into the network predictive structure model, the method further includes:
acquiring an initial text input by a user;
and carrying out syntactic/semantic processing on the initial text by using a word vector model to obtain the scoring information for representing the user car using experience.
As one implementation, the word vector model is a word2vec neural network used to construct the whole file.
As an implementation manner, the acquiring of the initial text input by the user specifically includes:
and acquiring the initial text from a user interactive webpage, a test database or a preset big data database.
As an embodiment, the step of establishing a network prediction structure model specifically includes:
and establishing a network prediction structure model with Bidirectional RNN Bidirectional recurrent neural network performance.
As one implementation mode, the structure of the bidirectional recurrent neural network comprises a timeout time step and rnn _ cell recurrent neural network basic units, and the knowledge map database adopts a neo4j map database.
As an embodiment, the step of adjusting the structural parameters of the network predictive structural model according to the loss function specifically includes:
minimizing the loss function to analyze function parameters causing a difference between the fuzzy intention and the standard intention;
and adjusting the structural parameters of the network prediction structural model according to the function parameters.
In order to solve the technical problem, the present application further provides a knowledge graph-based user vehicle usage experience investigation method, as one embodiment, the user vehicle usage experience investigation method adopts a classifier function constructed by the user vehicle usage experience investigation method, and the user vehicle usage experience investigation method includes the steps of:
Acquiring an initial text from the vehicle using experience feedback information of the user;
inputting the initial text into the classifier function for analysis to derive a true intent for expressing a user's car experience.
As one embodiment, the step of obtaining the initial text from the user car use experience feedback information specifically includes:
and acquiring an initial text from the vehicle using experience feedback information of the user by utilizing a text mining technology.
In order to solve the technical problem, the present application further provides a knowledge graph-based user car experience investigation system, and as an embodiment of the user car experience investigation system, the user car experience investigation system is configured with a processor, and the processor is configured to execute program data to implement the user car experience investigation method.
In order to solve the technical problem, the present application further provides a cloud server, and as an embodiment, the cloud server is configured with the above knowledge graph-based user vehicle experience research system.
The method, the system and the cloud server for the user vehicle using experience investigation based on the knowledge graph firstly establish a network prediction structure model, scoring information representing the user's vehicular experience is then input to the network predictive structure model, judging the fuzzy intention of the grading information through the network prediction structure model, inputting the grading information into a knowledge map database, judging the standard intention of the grading information by using the knowledge map database, then comparing the fuzzy intention and the standard intention, establishing a loss function of the scoring information according to the standard intention, and adjusting the structural parameters of the network prediction structure model according to the loss function, and updating the adjusted structural parameters to the network prediction structure model to obtain an expected classifier function for judging the actual user vehicle using experience intention. The method and the device can utilize the algorithm network to construct the judgment model, and parameters of the judgment model can be continuously corrected according to actual conditions, so that the grading can be automatically obtained, effective information of the real intention of a user can be intelligently identified according to the grading, development requirements can be made by research personnel according to the effective information, products can adapt to the development trend of market demands, and the product competitiveness of companies is improved.
The foregoing description is only an overview of the technical solutions of the present application, and in order to make the technical means of the present application more clearly understood, the present application may be implemented in accordance with the content of the description, and in order to make the above and other objects, features, and advantages of the present application more clearly understood, the following preferred embodiments are described in detail with reference to the accompanying drawings.
Drawings
Fig. 1 is a schematic flowchart illustrating an embodiment of a vehicle experience investigation method for a user.
Fig. 2 is a flowchart illustrating another embodiment of the vehicle using experience research method for the user.
FIG. 3 is a schematic diagram of an embodiment of a vehicle experience research system for a user.
Detailed Description
To further clarify the technical measures and effects taken by the present application to achieve the intended purpose, the present application will be described in detail below with reference to the accompanying drawings and preferred embodiments.
While the present application has been described in terms of specific embodiments and examples for achieving the desired objects and objectives, it is to be understood that the invention is not limited to the disclosed embodiments, but is to be accorded the widest scope consistent with the principles and novel features as defined by the appended claims.
Referring to fig. 1, fig. 1 is a schematic flow chart illustrating an embodiment of a method for researching a vehicle experience of a user.
As one embodiment, the method for researching the user's experience in the vehicle may include, but is not limited to, the following steps.
Step S101, a network prediction structure model is established;
step S102, inputting scoring information used for representing vehicle experience of a user into the network prediction structure model, and judging fuzzy intention of the scoring information through the network prediction structure model;
step S103, inputting the scoring information into a knowledge map database, and judging a standard intention of the scoring information by using the knowledge map database;
step S104, comparing the fuzzy intention with the standard intention, and establishing a loss function of the scoring information according to the standard intention;
and S105, adjusting the structural parameters of the network prediction structure model according to the loss function, and updating the adjusted structural parameters to the network prediction structure model to obtain an expected classifier function for judging the real user vehicle using experience intention.
In addition, before the step of inputting the scoring information representing the user experience with the vehicle into the network predictive structure model according to the present embodiment, the method may further include: acquiring an initial text input by a user; and carrying out syntactic/semantic processing on the initial text by using a word vector model to obtain the scoring information for representing the user car using experience.
Specifically, the word vector model in this embodiment is a word2vec neural network used for constructing an entire file.
It should be noted that the word2vec neural network in the embodiment may be a group of related models for generating word vectors, and the related models are shallow and double-layer neural networks for training to reconstruct the word text of linguistics. In the implementation process, input words at adjacent positions can be guessed, the order of words is not important under the assumption of a bag-of-words model in word2vec, and after training is completed, the word2vec model in the embodiment can be used for mapping each word to a vector and can be used for representing the relation between words.
It should be noted that, in this embodiment, the acquiring an initial text input by a user specifically includes: and acquiring the initial text from a user interactive webpage, a test database or a preset big data database.
It is understood that the web page interacted with the embodiment may be a dedicated web page for leaving messages and feeding back and evaluating for the user, or a comment web page of various chat software, such as a circle of friends. The test database may be a random database specially used for building a model for testing, and the content of the random database may be obtained from various webpages.
It should be noted that, the step of establishing the network prediction structure model according to this embodiment may specifically include: and establishing a network prediction structure model with the performance of the Bidirectional RNN.
Specifically, the structure of the bidirectional recurrent neural network comprises a time step and rnn _ cell recurrent neural network basic units, and the knowledge map database adopts a neo4j map database.
It should be noted that the neo4j graph database according to this embodiment may be a NoSql (non-relational) database written by Java language and Scala multi-paradigm programming language, and is specifically used for storing a network graph. By the method, the embodiment can perform rapid database operation, and the data is more visual and flexible.
The neo4j graph database according to this embodiment may use a special data storage structure and a specially optimized graph algorithm.
In addition, the step of adjusting the structural parameters of the network prediction structure model according to the loss function in the present embodiment may specifically include: minimizing the loss function to analyze function parameters causing a difference between the fuzzy intention and the standard intention; and adjusting the structural parameters of the network prediction structural model according to the function parameters.
The method and the device can utilize the algorithm network to construct the judgment model, and parameters of the judgment model can be continuously corrected according to actual conditions, so that the grading can be automatically obtained, effective information of the real intention of a user can be intelligently identified according to the grading, development requirements can be made by research personnel according to the effective information, products can adapt to the development trend of market demands, and the product competitiveness of companies is improved.
Referring to fig. 1 and fig. 2, fig. 2 is a schematic flow chart illustrating another embodiment of a user experience investigation method according to the present application.
It should be noted that, in the present embodiment, the user experience research method using car includes a classifier function constructed by the user experience research method using car as shown in fig. 1 and the embodiments thereof, and the user experience research method using car as shown in the present embodiment may include steps.
Step S201, acquiring an initial text from the user car use experience feedback information;
step S202, inputting the initial text into the classifier function to analyze and obtain a real intention for expressing the vehicle using experience of the user.
Further, the step of obtaining the initial text from the user car use experience feedback information in this embodiment may specifically include: and acquiring an initial text from the vehicle using experience feedback information of the user by utilizing a text mining technology.
For example, the processing procedure of the text mining technology of the present embodiment may include the following procedures.
1.1: creating a label feedback information base:
1.1.1: extracting a feedback information sample, cleaning the sample, and cleaning audio, video, pictures, incomplete feedback information, messy codes and illegal characters;
1.1.2: manually classifying according to a tag definition library;
1.1.3: carrying out dynamic clustering and fuzzy clustering on the samples at the same time, and setting cluster parameters;
1.1.4: carrying out semantic analysis, cluster feature analysis, cluster parameter correction and density noise reduction treatment in sequence to obtain a noise value M;
1.1.5: comparing the noise value M with a threshold value a, if the noise value M is smaller than the threshold value a, skipping to the step 1.1.6, and if the noise value M is larger than or equal to the threshold value a, skipping to the step 1.1.3;
1.1.6: then carrying out model clustering, semantic analysis, class characteristic analysis, class parameter correction and density noise reduction treatment in sequence to obtain a noise value N;
1.1.7: comparing the noise value N with a threshold value a, if the noise value N is smaller than the threshold value a, skipping to the step 1.1.8, and if the noise value N is larger than or equal to the threshold value a, skipping to the step 1.1.6 after correcting the label definition library;
1.1.8: carrying out model classification to form a label feedback information base;
1.2: creating a characteristic feedback information base:
1.2.1: carrying out sample word frequency analysis and semantic analysis on the tag feedback information base in sequence;
1.2.2: performing high word frequency classification;
1.2.3: creating a mapping model of the feature word and label definition library to form a feature feedback information library;
1.3: updating and maintaining a feedback information base:
1.3.1: extracting a full-amount classified feedback information sample;
1.3.2: performing word frequency analysis, semantic analysis, density noise reduction processing and noise data cleaning in sequence, classifying samples, and updating a label feedback information base or a characteristic feedback information base;
1.3.3: collecting new labels, extracting feedback information samples with the new labels, entering step 1.1, cleaning noise data, classifying samples, and updating a label feedback information base;
(2) acquiring a user vehicle experience feedback information set:
2.1: extracting a full-scale historical feedback information sample of an online user and an offline user, cleaning the sample, and cleaning video, audio and pictures;
2.2: carrying out dynamic clustering and fuzzy clustering synchronous processing on the samples, and then sequentially carrying out word frequency analysis, semantic analysis, class characteristic analysis, class parameter correction and density noise reduction processing to obtain a noise value A;
2.3: comparing the noise value A with a threshold value a, if the noise value A is smaller than the threshold value a, skipping to the step 2.4, and if the noise value A is larger than or equal to the threshold value a, skipping to the step 2.2;
2.4: then carrying out model clustering, semantic analysis, class feature analysis and density noise reduction treatment in sequence to obtain a noise value B;
2.5: comparing the noise value B with the threshold a, if the noise value B is smaller than the threshold a, skipping to the step 2.6, and if the noise value B is larger than or equal to the threshold a, skipping to the step 2.4 after correcting the parameters;
2.6: carrying out model classification to form a user vehicle experience feedback information set;
in the above process, the threshold a may be adjusted according to the service requirement; the definition of the following terms involved in the above process is as follows:
tag definition library: the method comprises the following steps that a class library formed by one class of self-defined labels is provided, each label points to objects with the same class attribute, and different classes of labels have obvious characteristic difference and follow the principles of high clustering and low coupling;
cluster parameters: when clustering is carried out by using a clustering algorithm, the number of the groups is set manually according to the number of the label types of the label definition library and the similarity of feedback information, the sample similarity of the same group is higher, the sample similarity of different groups is lower, the parameter is used as the grouping basis during clustering, and the parameter is continuously adjusted in a manual supervision mode so as to achieve the purpose of optimal matching with the label definition library;
Semantic analysis: first, manual analysis: after clustering samples, manually understanding the samples in a manual sampling mode, judging the process of similarity between the samples and simultaneously serving as a modification basis of cluster parameters; second, machine analysis: when samples are classified, the samples are classified through a matching algorithm with a feedback information base and are used as a basis for correcting the feedback information base;
analyzing cluster characteristics: performing a feature extraction and identification process on the clustered clusters by semantic analysis and an algorithm for extracting main features;
modifying cluster parameters: when a feedback information base is constructed, after a sample is clustered for the first time, the cluster feature analysis is utilized in a mode of artificial supervision and learning, the cluster group quantity is adjusted to achieve the best matching with a label definition base, and the process of adjusting the group quantity is a correction cluster parameter;
density denoising treatment: in the cluster feature analysis process, noise processing needs to be carried out on data, points with longer distances in a main feature scatter point distribution diagram are removed to form a category set capable of reflecting main features, and the process of removing noise points is density noise reduction processing;
Class feature analysis: performing a first cluster denoising, and performing a feature extraction and identification process on the denoised category set;
correcting the class parameters: when a feedback information base is constructed, after the samples are clustered for the second time, the group quantity of the clusters is adjusted to achieve the best matching with the label definition base by utilizing the class characteristic analysis in a mode of manual supervision and learning, and the process of adjusting the group quantity is a correction class parameter;
and (4) modifying the label definition library: in the process of the second clustering, because the noise reduction processing is carried out for one time, the sample classification model preliminarily meets the principles of high clustering and low coupling, and after the second noise reduction processing is carried out based on the model, the business requirements can be basically met, the classification model at the moment is determined, the optimal matching with the classification is achieved by adjusting the label definition library, and the adjusting process is a modified label definition library;
classifying based on a model: after two times of noise reduction, a classification model based on samples is formed and is used as a correction algorithm of cold start, and then the samples needing to be classified are classified based on the model;
dynamic clustering: finding sample vocabularies which accord with the categories according to the defined categories;
Fuzzy clustering: according to the sample vocabulary semantic fuzzy attribution category;
model clustering: a category is assumed, and then a sample vocabulary conforming to the category is found, so that the best fit between the given category and the sample vocabulary is achieved.
Referring to fig. 3, fig. 3 is a schematic structural diagram of an embodiment of a user experience research system.
The user car use experience research system according to the present embodiment is provided with a processor 31, and the processor 31 is configured to execute program data to implement the user car use experience research method.
Specifically, the processor 31 is configured to build a network prediction structure model;
the processor 31 is configured to input scoring information representing a user experience of using a vehicle into the network predictive structure model, and determine a fuzzy intention of the scoring information through the network predictive structure model;
the processor 31 is used for inputting the scoring information into a knowledge map database, and judging a standard intention of the scoring information by using the knowledge map database;
the processor 31 is used for comparing the fuzzy intention and the standard intention and establishing a loss function of the scoring information according to the standard intention;
The processor 31 is configured to adjust the structural parameters of the network predictive structural model according to the loss function, and update the adjusted structural parameters to the network predictive structural model, so as to obtain an expected classifier function for determining the real user experience intention.
It should be noted that, the processor 31 according to this embodiment may also be configured to obtain an initial text input by a user; and carrying out syntactic/semantic processing on the initial text by using a word vector model to obtain the scoring information for representing the user car using experience.
Specifically, the word vector model in this embodiment is a word2vec neural network used for constructing an entire file.
It should be noted that the word2vec neural network in the embodiment may be a group of related models for generating word vectors, and the related models are shallow and double-layer neural networks for training to reconstruct the word text of linguistics. In the implementation process, input words at adjacent positions can be guessed, the order of words is not important under the assumption of a bag-of-words model in word2vec, and after training is completed, the word2vec model in the embodiment can be used for mapping each word to a vector and can be used for representing the relation between words.
It should be noted that, the processor 31 according to this embodiment may be specifically configured to obtain the initial text from a web page interacted with by a user, a test database, or a preset big data database.
It is understood that the web page interacted with the embodiment may be a dedicated web page for leaving messages and feeding back and evaluating for the user, or a comment web page of various chat software, such as a circle of friends. The test database may be a random database specially used for building a model for testing, and the content of the random database may be obtained from various webpages.
It should be noted that, the network prediction structure model according to this embodiment may specifically be a network prediction structure model with Bidirectional RNN Bidirectional recurrent neural network performance.
Specifically, the structure of the bidirectional recurrent neural network comprises a time step and rnn _ cell recurrent neural network basic units, and the knowledge map database adopts a neo4j map database.
It should be noted that the neo4j graph database according to this embodiment may be a NoSql (non-relational) database written by Java language and Scala multi-paradigm programming language, and is specifically used for storing a network graph. By the method, the embodiment can perform rapid database operation, and the data is more visual and flexible.
The neo4j graph database according to this embodiment may use a special data storage structure and a specially optimized graph algorithm.
It should be noted that, the processor 31 of this embodiment may be specifically configured to perform a minimization process on the loss function to analyze and obtain a function parameter that causes a difference between the fuzzy intent and the standard intent; and adjusting the structural parameters of the network prediction structure model according to the function parameters.
In this embodiment, the processor 31 is configured to obtain an initial text from the user experience feedback information;
the processor 31 is configured to input the initial text into the classifier function for analysis to obtain a real intention for expressing the user's car experience.
Further, the processor 31 according to this embodiment may be specifically configured to obtain the initial text from the user car experience feedback information by using a text mining technology.
In order to solve the technical problem, the present application further provides a cloud server, as an embodiment, the cloud server is configured with the knowledge graph-based user experience research system described in fig. 3 and the embodiment thereof.
The method and the device can utilize the algorithm network to construct the judgment model, and parameters of the judgment model can be continuously corrected according to actual conditions, so that the grading can be automatically obtained, effective information of the real intention of a user can be intelligently identified according to the grading, development requirements can be made by research personnel according to the effective information, products can adapt to the development trend of market demands, and the product competitiveness of companies is improved.
The following description is briefly made in conjunction with a specific application example of the present application.
Fig. 1 and the training phase according to the embodiment:
1. inputting scoring information of a plurality of text comments;
2. word segmentation processing is carried out by using word2 vec;
3. defining Bi _ RNN (bidirectional recurrent neural network), wherein the structure of the Bi _ RNN comprises timekeep and RNN _ cell;
4. obtaining fuzzy intentions of classification results through the network prediction structure model, such as satisfaction, dissatisfaction and generality;
5. comparing the standard intention of the scoring information judged by the knowledge map database, and establishing an information loss function, wherein the knowledge map database is marked with true and false as a knowledge reference;
6. minimizing a loss function, and continuously adjusting the structure parameters of the network prediction structure model
7. A classifier function F is fitted.
Next in the prediction phase of fig. 2 and its embodiments:
1. inputting an initial text of any user car experience feedback information into a classifier function F;
2. obtaining a prediction classification: including unsatisfactory, satisfactory, or general.
It should be added that, in the embodiment, a knowledge map database with authenticity of judgment satisfaction needs to be constructed in advance; the satisfaction degree investigation of the user vehicle using experience can be understood as the satisfaction evaluation of the customer on the service or the commodity, and the satisfaction degree of the commodity is truly reflected.
In the present embodiment, the true intention may be understood as whether the intention of the user really exists or not, and a malicious intention (influencing factor) or a recorded false intention is excluded.
The method and the device can utilize the algorithm network to construct the judgment model, and parameters of the judgment model can be continuously corrected according to actual conditions, so that the grading can be automatically obtained, effective information of the real intention of a user can be intelligently identified according to the grading, development requirements can be made by research personnel according to the effective information, products can adapt to the development trend of market demands, and the product competitiveness of companies is improved.
Although the present application has been described with reference to a preferred embodiment, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the application, and all changes, substitutions and alterations that fall within the spirit and scope of the application are to be understood as being included within the following description of the preferred embodiment.
Claims (11)
1. A user vehicle using experience investigation method based on a knowledge graph is characterized by comprising the following steps:
establishing a network prediction structure model;
inputting scoring information used for representing vehicle experience of a user into the network prediction structure model, and judging fuzzy intention of the scoring information through the network prediction structure model;
inputting the scoring information into a knowledge graph database, and judging a standard intention of the scoring information by using the knowledge graph database;
comparing the fuzzy intention with the standard intention, and establishing a loss function of the scoring information according to the standard intention;
and adjusting the structural parameters of the network prediction structure model according to the loss function, and updating the adjusted structural parameters to the network prediction structure model to obtain an expected classifier function for judging the actual user vehicle using experience intention.
2. The method of claim 1, wherein the step of inputting scoring information representing the user's vehicular experience into the network predictive structure model further comprises:
acquiring an initial text input by a user;
And carrying out syntactic/semantic processing on the initial text by using a word vector model to obtain the scoring information for representing the user car using experience.
3. The user vehicular experience investigation method of claim 2, wherein the word vector model is a word2vec neural network used to construct an entire file.
4. The method according to claim 3, wherein the obtaining of the initial text input by the user specifically comprises:
and acquiring the initial text from a user interactive webpage, a test database or a preset big data database.
5. The method according to claim 3, wherein the step of building a network predictive structure model specifically comprises:
and establishing a network prediction structure model with the performance of the Bidirectional RNN.
6. The user experience investigation method of claim 5, wherein the structure of the bidirectional recurrent neural network comprises a time step and rnn _ cell recurrent neural network basic units, and the knowledge map database adopts a neo4j map database.
7. The method according to claim 5, wherein the step of adjusting the structure parameters of the network predictive structure model according to the loss function specifically comprises:
Minimizing the loss function to analyze function parameters causing a difference between the fuzzy intention and the standard intention;
and adjusting the structural parameters of the network prediction structure model according to the function parameters.
8. A method for user vehicular experience research based on knowledge graph, wherein the method for user vehicular experience research adopts a classifier function constructed by the method for user vehicular experience research according to any one of claims 1-7, and the method for user vehicular experience research comprises the following steps:
acquiring an initial text from the user vehicle experience feedback information;
inputting the initial text into the classifier function for analysis to derive a true intent for expressing a user's car experience.
9. The method of claim 8, wherein the step of obtaining the initial text from the user car use experience feedback information comprises:
and acquiring an initial text from the vehicle using experience feedback information of the user by utilizing a text mining technology.
10. A knowledge-graph-based user vehicular experience research system, wherein the user vehicular experience research system is configured with a processor for executing program data to implement the user vehicular experience research method according to any one of claims 1-9.
11. A cloud server, characterized in that the cloud server is configured with the knowledgegraph-based user in-vehicle experience research system of claim 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811146863.4A CN110968698B (en) | 2018-09-29 | 2018-09-29 | User vehicle using experience investigation method and system based on knowledge graph and cloud server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811146863.4A CN110968698B (en) | 2018-09-29 | 2018-09-29 | User vehicle using experience investigation method and system based on knowledge graph and cloud server |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110968698A CN110968698A (en) | 2020-04-07 |
| CN110968698B true CN110968698B (en) | 2022-07-29 |
Family
ID=70027566
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811146863.4A Active CN110968698B (en) | 2018-09-29 | 2018-09-29 | User vehicle using experience investigation method and system based on knowledge graph and cloud server |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110968698B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107451230A (en) * | 2017-07-24 | 2017-12-08 | 宗晖（上海）机器人有限公司 | A kind of answering method and question answering system |
| CN107729444A (en) * | 2017-09-30 | 2018-02-23 | 桂林电子科技大学 | Recommend method in a kind of personalized tourist attractions of knowledge based collection of illustrative plates |
| WO2018055646A1 (en) * | 2016-09-22 | 2018-03-29 | Dogma Srl. | Method and system for searching, publishing and managing the life cycle of multimedia contents related to public events and the user experience |
| CN107862561A (en) * | 2017-09-15 | 2018-03-30 | 广州唯品会研究院有限公司 | A kind of method and apparatus that user-interest library is established based on picture attribute extraction |
| CN107885760A (en) * | 2016-12-21 | 2018-04-06 | 桂林电子科技大学 | It is a kind of to represent learning method based on a variety of semantic knowledge mappings |
-
2018
- 2018-09-29 CN CN201811146863.4A patent/CN110968698B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018055646A1 (en) * | 2016-09-22 | 2018-03-29 | Dogma Srl. | Method and system for searching, publishing and managing the life cycle of multimedia contents related to public events and the user experience |
| CN107885760A (en) * | 2016-12-21 | 2018-04-06 | 桂林电子科技大学 | It is a kind of to represent learning method based on a variety of semantic knowledge mappings |
| CN107451230A (en) * | 2017-07-24 | 2017-12-08 | 宗晖（上海）机器人有限公司 | A kind of answering method and question answering system |
| CN107862561A (en) * | 2017-09-15 | 2018-03-30 | 广州唯品会研究院有限公司 | A kind of method and apparatus that user-interest library is established based on picture attribute extraction |
| CN107729444A (en) * | 2017-09-30 | 2018-02-23 | 桂林电子科技大学 | Recommend method in a kind of personalized tourist attractions of knowledge based collection of illustrative plates |
Non-Patent Citations (3)
| Title |
|---|
| 基于商品评论语义分析的情感知识图谱构建与查询应用;由丽萍等;《情报理论与实践》;20180315(第08期);全文 * |
| 电网公司客户服务知识图谱构建的应用价值;田晓等;《山东电力技术》;20151225(第12期);全文 * |
| 网络舆情倾向性预测模型仿真研究;连淑娟等;《计算机仿真》;20160315(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110968698A (en) | 2020-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114416927B (en) | Intelligent question-answering method, device, equipment and storage medium | |
| US20180357302A1 (en) | Method and device for processing a topic | |
| CN112632385A (en) | Course recommendation method and device, computer equipment and medium | |
| CN112836509A (en) | Expert system knowledge base construction method and system | |
| CN112417158A (en) | Training method, classification method, device and equipment of text data classification model | |
| CN104850617A (en) | Short text processing method and apparatus | |
| CN109710766B (en) | Complaint tendency analysis early warning method and device for work order data | |
| CN112418320B (en) | Enterprise association relation identification method, device and storage medium | |
| CN111931809A (en) | Data processing method and device, storage medium and electronic equipment | |
| CN110990563A (en) | Artificial intelligence-based traditional culture material library construction method and system | |
| CN110458600A (en) | Portrait model training method, device, computer equipment and storage medium | |
| CN111274390B (en) | Emotion cause determining method and device based on dialogue data | |
| CN110909768B (en) | Method and device for acquiring marked data | |
| CN104077408B (en) | Extensive across media data distributed semi content of supervision method for identifying and classifying and device | |
| CN113392920B (en) | Method, apparatus, device, medium, and program product for generating cheating prediction model | |
| CN110069558A (en) | Data analysing method and terminal device based on deep learning | |
| CN110968698B (en) | User vehicle using experience investigation method and system based on knowledge graph and cloud server | |
| CN110765872A (en) | Online mathematical education resource classification method based on visual features | |
| CN116958729A (en) | Training of object classification model, object classification method, device and storage medium | |
| CN113297482B (en) | User portrayal describing method and system of search engine data based on multiple models | |
| CN105138513A (en) | Method and device for determining similarity between Chinese vocabularies | |
| CN113822390B (en) | User portrait construction method and device, electronic equipment and storage medium | |
| CN115660695A (en) | Customer service personnel label portrait construction method and device, electronic equipment and storage medium | |
| US20220319504A1 (en) | Generating aspects from attributes identified in digital video audio tracks | |
| CN110472140B (en) | Object word recommendation method and device and electronic equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |