CN116739656B - Customer experience management method and system - Google Patents
Customer experience management method and system Download PDFInfo
- Publication number
- CN116739656B CN116739656B CN202311015467.9A CN202311015467A CN116739656B CN 116739656 B CN116739656 B CN 116739656B CN 202311015467 A CN202311015467 A CN 202311015467A CN 116739656 B CN116739656 B CN 116739656B
- Authority
- CN
- China
- Prior art keywords
- feature
- words
- word
- representing
- viewpoint
- 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
- 238000007726 management method Methods 0.000 title claims abstract description 20
- 238000000034 method Methods 0.000 claims abstract description 42
- 238000012545 processing Methods 0.000 claims abstract description 7
- 230000000717 retained effect Effects 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 53
- 238000000605 extraction Methods 0.000 claims description 46
- 239000013598 vector Substances 0.000 claims description 43
- 238000011156 evaluation Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 11
- 238000013528 artificial neural network Methods 0.000 claims description 9
- 230000011218 segmentation Effects 0.000 claims description 9
- 230000007246 mechanism Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 235000008694 Humulus lupulus Nutrition 0.000 claims 2
- 238000012163 sequencing technique Methods 0.000 claims 1
- 230000000379 polymerizing effect Effects 0.000 abstract 1
- 239000004973 liquid crystal related substance Substances 0.000 description 30
- 238000011835 investigation Methods 0.000 description 11
- 238000004364 calculation method Methods 0.000 description 6
- 238000007405 data analysis Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000008451 emotion Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000032683 aging Effects 0.000 description 2
- 230000003111 delayed effect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000003058 natural language processing Methods 0.000 description 2
- 206010063659 Aversion Diseases 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003607 modifier Substances 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
Classifications
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0203—Market surveys; Market polls
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses a client experience management method and a system, which belong to the technical field of data processing, wherein the method comprises the following steps: comment data of a target product on a plurality of platforms in a preset historical time period are obtained; extracting feature words and corresponding viewpoint words from the comment data; calculating the confidence coefficient of each feature word, and reserving the feature words with the confidence coefficient larger than the preset confidence coefficient and the corresponding viewpoint words; polymerizing the synonymous feature words to form a product feature cluster; counting the number of feature words contained in each product feature cluster, and reserving product feature clusters with the number of feature words larger than the preset number; determining the weight of the retained product feature cluster; calculating the data credibility of each platform; counting the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform; calculating customer experience satisfaction degree of a user on a target product; and when the customer experience satisfaction degree of the user on the target product is lower than the preset satisfaction degree, an alarm is sent out.
Description
Technical Field
The invention belongs to the technical field of data processing, and particularly relates to a client experience management method and system.
Background
In the development process of enterprises, good user public praise can bring more clients to the enterprises; conversely, poor user praise can cause an enterprise to lose an existing customer.
In the prior art, in order to obtain the public praise of the product of the enterprise, to know the approval degree and purchase intention of the product by the user, the enterprise often obtains the user survey data by sending a survey questionnaire to the user, and calculates a net recommendation value (NPS, net Promoter Score) according to the user survey data to measure the public praise of the user, the loyalty of the customer, and the like.
The existing workflow of net recommendation value acquisition mainly comprises: customer data before 1-2 months are manually extracted as investigated samples each month, NPS investigation is carried out on customers through a telephone mode, and then recommendation willingness scores of the customers and scoring reasons fed back by the customers are recorded into a system. And then, extracting a survey result data list from the system and feeding back the list to the NPS management department every month, and carrying out data cleaning, data analysis and the like by the NPS management department according to the list data and finally generating an NPS data analysis report. The whole process from the extraction of the initial client data to the acquisition of the NPS data analysis result has the problems of large workload of manual participation, unstable deviation of the extracted investigation sample, low calculation accuracy of the net recommended value, delayed investigation aging, difficulty in accurately evaluating the client experience and the like.
Disclosure of Invention
In order to solve the technical problems that in the prior art, user investigation data are obtained by sending a questionnaire to a user, and the net recommendation value is calculated according to the user investigation data to measure user public praise and the loyalty of a client, the workload of manual participation is large, time and labor are wasted, the extracted investigation sample is unstable in deviation, the net recommendation value is low in calculation accuracy, investigation aging is delayed, and accurate evaluation of client experience is difficult.
First aspect
The invention provides a client experience management method, which comprises the following steps:
s101: comment data of a target product on a plurality of platforms in a preset historical time period are obtained;
s102: extracting feature words and corresponding viewpoint words from the comment data through a syntactic dependency analysis algorithm;
s103: calculating the confidence coefficient of each feature word through a subject searching algorithm based on hyperlinks, and reserving the feature words with the confidence coefficient larger than the preset confidence coefficient and the corresponding viewpoint words;
s104: according to the semantic similarity among the feature words, the synonymous feature words are aggregated to form a product feature cluster;
s105: counting the number of feature words contained in each product feature cluster, reserving product feature clusters with the number of the feature words being larger than the preset number, and removing product feature clusters with the number of the feature words being smaller than or equal to the preset number;
s106: determining the weight of the retained product feature cluster;
s107: calculating the data credibility of each platform;
s108: counting the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform;
s109: calculating the customer experience satisfaction degree of a user on the target product according to the weight of each product feature cluster, the data credibility of each platform and the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform;
s110: and when the customer experience satisfaction degree of the user on the target product is lower than the preset satisfaction degree, an alarm is sent out.
Second aspect
The present invention provides a customer experience management system for executing the customer experience management method in the first aspect.
Compared with the prior art, the invention has at least the following beneficial technical effects:
(1) According to the method and the system, comment data of a target product on a plurality of platforms in a preset historical time period are obtained, feature words and corresponding viewpoint words are extracted according to the comment data on the plurality of platforms, the feature words are clustered into product feature clusters, and customer experience satisfaction of a user on the target product is automatically calculated according to the weight of each product feature cluster, the data credibility of each platform and the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and originate from each platform, without manual participation, time and labor are saved, data sources are stable, the customer experience satisfaction calculation accuracy is high, investigation timeliness is high, and customer experience on the target product can be accurately evaluated.
(2) In the invention, the confidence coefficient of the feature words is calculated by using the subject searching algorithm based on the hyperlink, and the feature words with the confidence coefficient larger than the preset threshold value and the corresponding viewpoint words are reserved, so that the reliability of data can be improved, unreliable comments and viewpoints are eliminated, and the experience and the opinion of a user are reflected more accurately.
(3) According to the invention, the synonymous feature words are aggregated into the product feature clusters according to the semantic similarity among the feature words, so that similar evaluation and opinion can be aggregated together, data redundancy is reduced, and more important feature information is extracted.
(4) In the invention, the data credibility of each platform is considered in the process of calculating the customer experience satisfaction, so that the malicious bad evaluation of the user on the individual platform is avoided, the data credibility is improved, and the user experience and opinion are reflected more accurately.
Drawings
The above features, technical features, advantages and implementation of the present invention will be further described in the following description of preferred embodiments with reference to the accompanying drawings in a clear and easily understood manner.
FIG. 1 is a schematic flow chart of a method for managing customer experience provided by the invention;
fig. 2 is a schematic structural diagram of a client experience management method provided by the present invention.
Detailed Description
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following description will explain the specific embodiments of the present invention with reference to the accompanying drawings. It is evident that the drawings in the following description are only examples of the invention, from which other drawings and other embodiments can be obtained by a person skilled in the art without inventive effort.
For simplicity of the drawing, only the parts relevant to the invention are schematically shown in each drawing, and they do not represent the actual structure thereof as a product. Additionally, in order to simplify the drawing for ease of understanding, components having the same structure or function in some of the drawings are shown schematically with only one of them, or only one of them is labeled. Herein, "a" means not only "only this one" but also "more than one" case.
It should be further understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.
In this context, it should be noted that the terms "mounted," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected, unless otherwise explicitly stated and defined. Either mechanically or electrically. Can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
In addition, in the description of the present invention, the terms "first," "second," and the like are used merely to distinguish between descriptions and are not to be construed as indicating or implying relative importance.
Example 1
In one embodiment, referring to fig. 1 of the specification, a schematic flow chart of a client experience management method provided by the present invention is shown. Referring to fig. 2 of the specification, a schematic structural diagram of a client experience management method provided by the invention is shown.
The invention provides a client experience management method, which comprises the following steps:
s101: and acquiring comment data of the target product on a plurality of platforms within a preset historical time period.
Alternatively, the target product may be a store, a restaurant, an application, a game, and the invention is not limited to the specific type of target product.
The size of the preset historical time period can be set by a person skilled in the art according to practical situations, and the invention is not limited.
Optionally, the preset history period is within the past month.
S102: and extracting feature words and corresponding viewpoint words from the comment data through a syntactic dependency analysis algorithm.
The syntactic dependency analysis is an important task in natural language processing, and aims to analyze syntactic dependency relationship among words in sentences and construct a dependency relationship tree or a dependency graph of the sentences. In the dependency tree or the dependency graph, each word serves as a node, and is connected to each other through a dependency relationship. Dependencies describe the modifiers between words, dependencies, and the grammatical structure of sentences.
Wherein, the feature words refer to words used for describing the features, attributes, functions and the like of the target product or service in the comments. For example, for a cell phone product, the feature words may include "screen size", "battery life", "performance", "design", and the like.
The viewpoint words are words used for expressing attitudes, emotions, evaluations or opinions of the features described by the feature words in the comments. For example, for the feature word "battery life" in a cell phone product, the terms may include positive terms such as "long", "durable", and negative terms such as "short", and the like.
In particular, predefined grammar rules and dependency patterns can be used to analyze the structure of sentences. Rule-based methods typically utilize pre-processing information such as part-of-speech tags, syntactic tags, etc., and apply a series of rules to determine dependencies between words.
In the practical application process, the applicant finds that the semantics of sentences are often formed by complex relationships and dependency relationships among different words, however, some sentences have complex grammar structures, the dependency relationships among words can span long distances and are difficult to capture accurately, and at the same time, some words can have ambiguity and can have different meanings in different contexts, which also leads to word segmentation difficulty. Thus, the applicant provides a brand new segmentation algorithm based on an attention mechanism.
In one possible implementation, S102 specifically includes substeps S1021 through S1026:
s1021: syntactical dependencies of the comment data are analyzed.
Wherein the syntactic dependency includes: the main-term relationship SBV, the centering relationship ATT, the in-form relationship ADV, the moving object relationship VOB, the parallel relationship CONJ and the like.
In particular, the syntactic dependencies may be analyzed using predefined grammar rules and dependency patterns, or may be analyzed using machine learning algorithms.
S1022: constructing a syntactic relation matrix according to syntactic dependency relations among all characters in comment data：
；
Wherein, the liquid crystal display device comprises a liquid crystal display device,representing syntactic relation matricesMMiddle (f)iLine 1jColumn element->Represent the firstiThe first characterjWhether there is a syntactic dependency between the individual characters, when +.>When=1, the first expressioniThe first characterjThere is a syntactic dependency between the individual characters when +.>When=0, the first expressioniThe first characterjThere is no syntactic dependency between the individual characters, +.>Represent the firstjRepresented node of the individual character, +.>Represent the firstiA set of character connected nodes,>represent the firstjThe characters communicate with the node set.
S1023: according to the syntactic relation matrixMAnd the positions of the characters in comment data to construct a position relation matrix：
；
Wherein, the liquid crystal display device comprises a liquid crystal display device,representing a positional relationship matrixPMiddle (f)iLine 1jColumn element->The representation is from the firstiThe first character to be dominatedjNumber of forward jumps of individual character, +.>Representing the slave dominant firstiThe first character tojReverse hop count of the individual characters.
S1024: introducing an attention mechanism, constructing a word segmentation neural network, and enabling a forward propagation process of the word segmentation neural network to be expressed as follows:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,represent the firstiOutput vector of individual characters, ">Represent the firstiPersonal wordSymbol and the firstjAttention weight between characters, +.>Represent the firstiFeature vector of individual character,/>Represent the firstjFeature vector of individual character,/>And->Model parameters representing a word segmentation neural network, +.>Representing an activation function->Represent the firstiThe first characterjThe relevance score between the individual characters,drepresenting the dimension of the feature vector ∈>Representing the matrix transpose.
It should be noted that the introduction of the attention mechanism allows the network to dynamically adjust the attention degree to different locations according to the input information. In particular, the attention mechanism may weight individual characters differently according to their relevance scores, thereby focusing more on important characters in a particular task, ignoring irrelevant characters. The word segmentation neural network that introduces a mechanism of attention may consider semantic relevance between words to better understand the structure and meaning of sentences. This helps to improve the accuracy of feature words and perspective words and avoid misunderstanding of irrelevant information.
S1025: and dividing the comment data into words according to the output vectors of the characters.
It should be noted that, by adopting the machine learning algorithm and the neural network, the whole process can realize automatic processing without manual intervention. Therefore, the efficiency and accuracy of processing comment data can be greatly improved, and the labor cost is saved.
Furthermore, the syntactic dependency analysis and the neural network method can support the processing of large-scale comment data, are suitable for comprehensively analyzing the evaluation of a large number of users on products or services, and help enterprises to better know the demands of the users and improve the products.
S1026: and extracting the characteristic words and the corresponding viewpoint words from the comment data according to various extraction rules.
In one possible implementation, S1026 specifically includes:
when the two words are the main predicate relationship SBV, the main predicate is taken as a characteristic word, and the predicate is taken as a viewpoint word corresponding to the characteristic word.
It should be noted that when two words are the main term relationship SBV: the main-predicate relationship means that one word is used as a subject, and the other word is used as a predicate, and an action executed by one subject is expressed. In this case, the subject is a feature word, and the predicate is a viewpoint word corresponding to the feature word. For example, for the sentence "a product is liked by a small mine", "a small mine" is a feature word, and "like" is a corresponding viewpoint word, the degree of like of a small mine for a product is described.
When the two words are the centering relation ATT, the center word is used as a characteristic word, and the fixed language is used as a viewpoint word corresponding to the characteristic word.
It should be noted that, when two words are the centering relationship ATT: centering relationship means that one word is used as a center word and the other word is used as a centering word for modifying the center word. In this case, the center word is used as a feature word, and the stationary word is used as a viewpoint word corresponding thereto. For example, for the sentence "this cell phone is large in screen", "cell phone" is a feature word, and "large in screen" is a corresponding viewpoint word, the feature of the cell phone is described.
When the two words are in the relationship ADV in terms of the word, the center word is used as a feature word, and the word is used as a viewpoint word corresponding to the feature word.
Note that, when two words are the in-word relationship ADV: the relationship in the form means that one word is used as a central word, and the other word is used as a scholarly word for describing the state or degree of the central word. In this case, the center word is used as a feature word, and the scholarly word is used as a viewpoint word corresponding thereto. For example, for the sentence "the service of this restaurant is very good", "restaurant" is a feature word, and "very good" is a corresponding viewpoint word, the service state of the restaurant is described.
Optionally, the extraction rule may further include:
when two words are a move guest relationship VOB: the guest-to-animal relationship means that one word is used as a verb and the other word is used as an object for describing an execution object of an action. In this case, the object may be a feature word and the verb may be a corresponding viewpoint word;
when two words are in a juxtaposition relationship CONJ: parallel relationship means that two words appear in parallel in the same sentence without a subordinate relationship. In this case, both of the parallel words may be used as feature words, but it is noted that there may be different viewpoints therebetween.
In the invention, the attention mechanism is introduced to enable the model to pay attention to key information in sentences more pertinently, and the long-distance dependency relationship can be captured better when the sentences with complex grammar structures are faced, so that the analysis result is more accurate. Further, in the face of words with ambiguity, the attention mechanism can help the model dynamically select the appropriate meaning according to context, thereby better understanding the semantics of the sentence.
S103: and calculating the confidence coefficient of each feature word through a subject searching algorithm based on hyperlinks, and reserving the feature words with the confidence coefficient larger than the preset confidence coefficient and the corresponding viewpoint words.
Among them, the hyperlink-based topic search algorithm is a method for calculating confidence of feature words, which can help identify important feature words and corresponding viewpoint words. The algorithm is based on the concept of a network graph, and uses the structure of hyperlinks to represent the relationship between feature words and viewpoint words, similar to the link relationship between web pages.
The size of the preset confidence coefficient can be set by a person skilled in the art according to practical situations, and the invention is not limited.
It should be noted that, the feature words with the confidence degree larger than the preset confidence degree and the corresponding viewpoint words are reserved, so that the reliability of the data can be increased, and redundant information in the data can be reduced.
In one possible embodiment, S103 specifically includes substeps S1031 to S1037:
s1031: defining the extraction rule as Hub node and the feature word as Authority nodeA bipartite directed graph representing feature words and extraction rules,Fa set of feature words is represented and,EXrepresenting a set of extraction rules>Representing that the extraction rule set points to the edge set of the feature word set, then the adjacency matrix of the bipartite directed graph between feature word and extraction rule +.>The method comprises the following steps:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,representing extraction rulesexAnd feature words->Relationships between (if->Is true and indicates the extraction ruleexCan successfully extract the characteristic words ++>，/>Is 1.
S1032: according to the characteristic words and the extractionAdjacency matrix taking bipartite directed graphs between rulesCalculating a confidence vector of the feature words and a confidence vector of the extraction rule:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,confidence vector representing feature words, ++>A confidence vector representing the extraction rule,representing the matrix transpose.
S1033: the viewpoint is taken as Hub node, the feature is taken as Authority node, and the definition is definedA bipartite directed graph representing the space between feature words and perspective words,Fa set of feature words is represented and,Orepresenting a set of ideological words,/->Representing that the set of perspective words points to the set of edges of the set of feature words, then the adjacency matrix of the bipartite directed graph between feature words and perspective words +.>The method comprises the following steps:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,representing viewpoint wordsoAnd feature words->Relationships between (if->Is true and represents a perspective wordoAnd feature words->At the same time present->Is 1.
S1034: adjacency matrix based on bipartite directed graph between feature words and perspective wordsCalculating a confidence vector of the feature word and a confidence vector of the viewpoint word:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,confidence vector representing feature words, ++>A confidence vector representing the feature word.
S1035: defining the extraction rule as Hub node and the viewpoint as Authority nodeA bipartite directed graph representing feature words and extraction rules,Oa set of perspective words is represented and,EXrepresenting a set of extraction rules>Representing that the extraction rule set points to the edge set of the viewpoint word set, then the adjacency matrix of the bipartite directed graph between the viewpoint word and the extraction rule +.>The method comprises the following steps:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,representing by extraction rulesexIdeographic wordsoRelationships between (if->Is true and indicates the extraction ruleexCan successfully extract the viewpoint wordso，/>Is 1.
S1036: calculating a confidence matrix of the extraction rule, the feature words and the viewpoint words according to the following formula:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,Aa confidence matrix representing the feature words,confidence matrix representing perspective words, ++>Confidence matrix representing extraction rules, +.>Representing the quality confidence parameter.
Wherein, the person skilled in the art can set the quality reliability parameter according to the actual situationThe size of (3) is not limited in the present invention.
The confidence matrix of the feature word is describedAConfidence matrix for viewpointConfidence of extraction ruleMatrix->The three mutually influence each other, and the confidence coefficient matrix of the feature wordsAConfidence matrix from viewpoint->And confidence matrix of extraction rule +.>Co-determination, confidence matrix of viewpoint>Confidence matrix of feature wordsAAnd confidence matrix of extraction rule +.>Co-decision, confidence matrix of extraction rule +.>Confidence matrix of feature wordsAConfidence matrix of viewpoint>And (5) jointly determining. The three mutually enhance the mutual credibility through the formula.
In the invention, by constructing the bipartite directed graph and the adjacency matrix, the relationship among the feature words, the extraction rules and the viewpoint words is comprehensively considered instead of being independently processed, so that more comprehensive and accurate information is obtained. By mutually enhancing the reliability, errors due to erroneous or incomplete information can be reduced. The selection of feature words and viewpoint words is critical to the satisfaction of subsequent computing customer experience, and erroneous judgment and erroneous results can be reduced by enhancing the credibility.
S1037: confidence matrix for feature wordsAThe confidence degrees of the feature words are ranked, and feature words with the confidence degrees larger than the preset confidence degrees and corresponding viewpoint words are reserved.
In the invention, the confidence coefficient of the feature words is calculated by using the subject searching algorithm based on the hyperlink, and the feature words with the confidence coefficient larger than the preset threshold value and the corresponding viewpoint words are reserved, so that the reliability of data can be improved, unreliable comments and viewpoints are eliminated, and the experience and the opinion of a user are reflected more accurately.
S104: and according to the semantic similarity among the feature words, the synonymous feature words are aggregated to form a product feature cluster.
In one possible implementation, S104 specifically includes substeps S1041 to S1045:
s1041: extracting word vectors of the feature words through a word embedding technology.
The word embedding technology is an important technology in the field of natural language processing, maps words in text into real vectors, and reserves semantic relations among the words in a vector space.
Specifically, the Word embedding technique may be Word2Vec, gloVe, fastText or the like.
S1042: random initializationKAnd clustering centers.
S1043: according to the word vector of each feature word, calculating the distance from the current feature word to the center point of each clustering center:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,representing the current feature word to the firstjDistance of individual cluster centers,/>The first of the word vectors representing the current feature wordiElement(s)>Represent the firstjThe first word vector of the center point of each cluster centeriThe number of elements to be added to the composition,srepresenting the dimension of the word vector.
S1044: dividing current feature words intoAnd in the smallest cluster, updating the cluster center.
S1045: continuing to select the next feature word until the clustering of all feature words is completed, and obtainingKAnd (3) product feature clusters.
According to the invention, the synonymous feature words are aggregated into the product feature clusters according to the semantic similarity among the feature words, so that similar evaluation and opinion can be aggregated together, data redundancy is reduced, and more important feature information is extracted.
S105: counting the number of feature words contained in each product feature cluster, reserving product feature clusters with the number of the feature words being larger than the preset number, and removing the product feature clusters with the number of the feature words being smaller than or equal to the preset number.
It should be noted that, by reserving product feature clusters with the number of feature words being greater than the preset number, clusters with the number of feature words being less than or equal to the preset number are removed, so that the accuracy of clustering can be improved, feature redundancy is reduced, computational complexity is reduced, more effective and accurate feature information is provided for subsequent data analysis and mining tasks, and understanding and insight of product features are enhanced.
S106: the weights of the clusters of retained product features are determined.
In one possible embodiment, S106 specifically includes substeps S1061 to S1063:
s1061: by comparing the characteristic clusters of each product in pairs and combining a nine-level scale method, a discrimination matrix is establishedA：
；
Wherein, the liquid crystal display device comprises a liquid crystal display device,represent the firstiThe product feature cluster is relative to the firstjImportance of individual product feature clusters, +.>The value of (c) can be determined by nine-pole scaling,nrepresenting the total number of product feature clusters.
The nine-pole scale method is a common quantitative evaluation method for comparing and judging the merits among a plurality of options or factors. And a nine-level scale method is used for enabling a decision maker to compare the importance between the product feature cluster A and the product feature cluster B. The result of the comparison is typically based on subjective perception, and the decision maker needs to select a number in a nine-level scale, representing the degree of importance of a relative to B.
S1062: calculating a discrimination matrixAFeature vectors and feature values of (a):
；
wherein, the liquid crystal display device comprises a liquid crystal display device,representing a discriminant matrixAIs used for the characteristic value of the (c),wrepresenting a discriminant matrixATakes the maximum characteristic value as +.>The feature vector corresponding to the largest feature value is denoted +.>，。
S1063: for the feature vector corresponding to the largest feature valueNormalization processing:
；
wherein the normalized vectorAre>Weights respectively representing the characteristic clusters of each product can be respectively marked as +.>。
It should be noted that, by determining the weight of the product feature cluster, the importance of the feature cluster can be quantified, help decision making, simplify the decision making process, and optimize product design and improvement. By doing so, product management and decision making can be more scientific and objective, and more valuable decision support is provided for enterprises.
S107: and calculating the data credibility of each platform.
In one possible embodiment, S107 specifically includes substeps S1071 to S1073:
s1071: and counting the number of feature words from each platform contained in each product feature cluster.
S1072: calculate the firstjProximity of comment data of each platform to other platforms:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,represent the firstjProximity of comment data of individual platform to other platforms, < ->Represent the firstiThe product feature clusters are derived fromjNumber of feature words of the individual platform, +.>Represent the firstiThe average number of feature words contained in each product feature cluster in each platform,nrepresenting the total number of product feature clusters.
It can be understood that if the proximity of comment data of a certain platform to comment data of other platforms is higher, the data provided by the platform has higher consistency and stability in the overall comment data, and the lower the possibility that the data of the platform has malicious poorly comment or malicious bill-refreshing behavior, the higher the data reliability of the data.
S1073: the proximity of each platform is standardized to obtain the firstjData credibility of each platform:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,represent the firstjThe degree of data trustworthiness of the individual platforms,mrepresenting the number of platforms.
In the invention, the data credibility of each platform is considered in the process of calculating the customer experience satisfaction, so that the malicious bad evaluation of the user on the individual platform is avoided, the data credibility is improved, and the user experience and opinion are reflected more accurately.
S108: and counting the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform.
Wherein, forward viewpoint words are used to represent attitudes of being positive, appreciating or supporting, and are generally used to express positive emotions of being good, satisfying, loving, etc. to something. For example: the words "excellent", "stick", "happy", "satisfactory" and the like all belong to forward viewpoint words.
Wherein, negative viewpoint words are used for expressing negative, dissatisfaction or criticizing attitudes, and are generally used for expressing negative emotions such as dissatisfaction, aversion, negative evaluation and the like of something. For example: words such as "bad", "open", "disappointed" and "offensive" are negative terms.
S109: and calculating the customer experience satisfaction degree of the user on the target product according to the weight of each product feature cluster, the data credibility of each platform and the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform.
In one possible implementation, S109 is specifically:
calculating the customer experience satisfaction degree of the user for the target product according to the following formula:
；
wherein, the liquid crystal display device comprises a liquid crystal display device,Sindicating the satisfaction of the customer experience,represent the firstiWeights of individual product feature clusters, +.>Represent the firstjData credibility of individual platforms, +.>Represent the firstiThe product feature clusters are derived fromjNumber of forward views of individual platforms, +.>Represent the firstiThe product feature clusters are derived fromjThe number of negative-going terms of opinion of the individual platform,nrepresenting the number of clusters of features of the product,mrepresenting the number of platforms.
According to the method and the system, the customer experience satisfaction degree of the user on the target product is automatically calculated according to the weight of each product feature cluster, the data credibility of each platform and the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform, so that the method and the system are free of manual participation, time and labor are saved, the data sources are stable, the calculation accuracy of the customer experience satisfaction degree is high, the investigation timeliness is high, and the customer experience on the target product can be accurately evaluated.
S110: and when the customer experience satisfaction degree of the user on the target product is lower than the preset satisfaction degree, an alarm is sent out.
The size of the preset satisfaction degree can be set by a person skilled in the art according to practical situations, and the invention is not limited.
In one possible implementation, S110 is specifically:
when the customer experience satisfaction of the user for the target product is lower than the preset satisfaction, an alarm is sent out in at least one of the following modes:
and displaying an alarm popup window.
An alert mail is sent to notify the relevant personnel.
Or sending an alarm short message to inform related personnel.
It should be noted that, sending out an alarm is an important means for responding to user dissatisfaction in time, which is helpful to improve product quality and user experience, and enhance competitive power and reputation of enterprises. Meanwhile, the setting of the alarm system needs to be determined according to actual conditions and preset satisfaction, so that the accuracy and the credibility of the alarm are ensured.
Compared with the prior art, the invention has at least the following beneficial technical effects:
(1) According to the method and the system, comment data of a target product on a plurality of platforms in a preset historical time period are obtained, feature words and corresponding viewpoint words are extracted according to the comment data on the plurality of platforms, the feature words are clustered into product feature clusters, and customer experience satisfaction of a user on the target product is automatically calculated according to the weight of each product feature cluster, the data credibility of each platform and the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and originate from each platform, without manual participation, time and labor are saved, data sources are stable, the customer experience satisfaction calculation accuracy is high, investigation timeliness is high, and customer experience on the target product can be accurately evaluated.
(2) In the invention, the confidence coefficient of the feature words is calculated by using the subject searching algorithm based on the hyperlink, and the feature words with the confidence coefficient larger than the preset threshold value and the corresponding viewpoint words are reserved, so that the reliability of data can be improved, unreliable comments and viewpoints are eliminated, and the experience and the opinion of a user are reflected more accurately.
(3) According to the invention, the synonymous feature words are aggregated into the product feature clusters according to the semantic similarity among the feature words, so that similar evaluation and opinion can be aggregated together, data redundancy is reduced, and more important feature information is extracted.
(4) In the invention, the data credibility of each platform is considered in the process of calculating the customer experience satisfaction, so that the malicious bad evaluation of the user on the individual platform is avoided, the data credibility is improved, and the user experience and opinion are reflected more accurately.
Example 2
In one embodiment, the present invention provides a client experience management system for executing the client experience management method in embodiment 1.
The steps and effects of the client experience management method in the above embodiment 1 can be realized by the client experience management system provided by the present invention, and in order to avoid repetition, the present invention is not repeated.
Compared with the prior art, the invention has at least the following beneficial technical effects:
(1) According to the method and the system, comment data of a target product on a plurality of platforms in a preset historical time period are obtained, feature words and corresponding viewpoint words are extracted according to the comment data on the plurality of platforms, the feature words are clustered into product feature clusters, and customer experience satisfaction of a user on the target product is automatically calculated according to the weight of each product feature cluster, the data credibility of each platform and the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and originate from each platform, without manual participation, time and labor are saved, data sources are stable, the customer experience satisfaction calculation accuracy is high, investigation timeliness is high, and customer experience on the target product can be accurately evaluated.
(2) In the invention, the confidence coefficient of the feature words is calculated by using the subject searching algorithm based on the hyperlink, and the feature words with the confidence coefficient larger than the preset threshold value and the corresponding viewpoint words are reserved, so that the reliability of data can be improved, unreliable comments and viewpoints are eliminated, and the experience and the opinion of a user are reflected more accurately.
(3) According to the invention, the synonymous feature words are aggregated into the product feature clusters according to the semantic similarity among the feature words, so that similar evaluation and opinion can be aggregated together, data redundancy is reduced, and more important feature information is extracted.
(4) In the invention, the data credibility of each platform is considered in the process of calculating the customer experience satisfaction, so that the malicious bad evaluation of the user on the individual platform is avoided, the data credibility is improved, and the user experience and opinion are reflected more accurately.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the invention, which are described in detail and are not to be construed as limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (9)
1. A method for managing customer experience, comprising:
s101: comment data of a target product on a plurality of platforms in a preset historical time period are obtained;
s102: extracting feature words and corresponding viewpoint words from the comment data through a syntactic dependency analysis algorithm;
s103: calculating the confidence coefficient of each feature word through a subject searching algorithm based on hyperlinks, and reserving the feature words with the confidence coefficient larger than the preset confidence coefficient and the corresponding viewpoint words;
s104: according to the semantic similarity among the feature words, the synonymous feature words are aggregated to form a product feature cluster;
s105: counting the number of feature words contained in each product feature cluster, reserving product feature clusters with the number of the feature words being larger than the preset number, and removing product feature clusters with the number of the feature words being smaller than or equal to the preset number;
s106: determining the weight of the retained product feature cluster;
s107: calculating the data credibility of each platform;
s108: counting the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform;
s109: calculating the customer experience satisfaction degree of a user on the target product according to the weight of each product feature cluster, the data credibility of each platform and the number of positive viewpoint words and negative viewpoint words which are contained in each product feature cluster and are derived from each platform;
s110: when the customer experience satisfaction degree of the user for the target product is lower than the preset satisfaction degree, an alarm is sent;
wherein, the step S103 specifically includes:
s1031: defining G by taking the extraction rule as Hub node and the feature word as Authority node 1 ＝(F,EX,E 1 ) Representing a bipartite directed graph between feature words and extraction rules, F representing a feature word set, EX representing an extraction rule set, E 1 Representing that the extraction rule set points to the edge set of the feature word set, then the feature word
Adjacency matrix M of bipartite directed graph between extraction rules F_E The method comprises the following steps:
wherein (f, ex) represents the relation between the extraction rule ex and the feature word f, if (f, ex) E E 1 Is true, meaning that the extraction rule ex can successfully extract the feature words f, M F_E 1 is counted;
s1032: adjacency matrix M of bipartite directed graph between feature words and extraction rules F_E Calculating confidence vector of feature words and extraction ruleConfidence vector:
wherein A is F_E Confidence vector representing feature words, H F_E Confidence vector representing extraction rule, (·) T Representing a matrix transpose;
s1033: defining G by taking viewpoint words as Hub nodes and feature words as Authority nodes 2 = (F, O, E) represents a bipartite directed graph between feature words and perspective words, F represents a feature word set, O represents a perspective word set, E 2 Representing that the viewpoint word set points to the edge set of the feature word set, then the adjacency matrix M of the bipartite directed graph between the feature word and the viewpoint word F_O The method comprises the following steps:
wherein (o, f) represents the relationship between the viewpoint word o and the feature word f, if (o, f) ∈E 2 And (3) if true, representing that the viewpoint word o and the feature word f appear simultaneously, M F_O 1 is counted;
s1034: adjacency matrix M of bipartite directed graph between feature words and viewpoint words F_O Calculating a confidence vector of the feature word and a confidence vector of the viewpoint word:
wherein A is F_O Confidence vector representing feature words, H F_O A confidence vector representing the feature word;
s1035: defining G by taking the extraction rule as Hub node and the viewpoint as Authority node 3 ＝(O,EX,E 3 ) Represents a bipartite directed graph between feature words and extraction rules, O represents a set of perspective words, EX represents a set of extraction rules, E 3 Representing extraction rule set pointing perspectiveEdge set of word set, adjacency matrix M of bipartite directed graph between viewpoint words and extraction rule O_E The method comprises the following steps:
wherein (o, ex) represents the relation between the rule ex and the viewpoint o by extraction, if (o, ex) E E 3 Is true, meaning that the extraction rule ex can successfully extract the viewpoint o, M O_E 1 is counted;
s1036: calculating a confidence matrix of the extraction rule, the feature words and the viewpoint words according to the following formula:
wherein A represents a confidence matrix of the feature words, H O Confidence matrix, H, representing perspective words E A confidence matrix representing the extraction rule, λ representing a quality confidence parameter;
s1037: and sequencing the confidence coefficient of each feature word in the confidence coefficient matrix A of the feature word, and reserving the feature words with the confidence coefficient larger than the preset confidence coefficient and the corresponding viewpoint words.
2. The method for managing customer experience according to claim 1, wherein S102 specifically comprises:
s1021: analyzing syntactic dependencies of the comment data;
s1022: constructing a syntactic relation matrix M= [ M ] according to syntactic dependency relation among all characters in the evaluation data ij ]：
Wherein m is ij Elements representing the jth column of the ith row in the syntactic relationship matrix M, M ij Represent the firstWhether there is a syntactic dependency between the i character and the j-th character, when m ij When=1, it indicates that there is a syntactic dependency between the ith character and the jth character, when m ij When=0, it indicates that there is no syntactic dependency between the ith character and the jth character, c j Represented node representing the jth character, S i Represents the i-th character connected node set S j Representing a j-th character connected node set;
s1023: constructing a position relation matrix P= [ P ] according to the syntactic relation matrix M and the positions of the characters in the comment data ij ]：
Wherein p is ij Elements representing the j-th column of the i-th row in the positional relationship matrix P, l ij Indicating the number of forward hops from the ith character to the j-th character that is dominant,indicating the number of reverse hops from the ith character to the jth character that are dominant;
s1024: introducing an attention mechanism, constructing a word segmentation neural network, wherein the forward propagation process of the word segmentation neural network can be expressed as follows:
wherein o is i Output vector representing the ith character, a ij Represents the attention weight between the ith character and the jth character, h i Feature vector, h, representing the ith character j Feature vector representing jth character, W 1 And W is 2 Model parameters representing the word-segmentation neural network, reLU (·) representing the activation function, e ij Represents the relevance score between the ith and jth characters, d represents the dimension of the feature vectorDegree (·) T Representing a matrix transpose;
s1025: according to the output vector of each character, word segmentation is carried out on the evaluation data;
s1026: and extracting feature words and corresponding viewpoint words from the comment data according to various extraction rules.
3. The method for managing customer experience according to claim 2, wherein S1026 specifically comprises:
when the two words are main-predicate relation SBV, the main words are taken as characteristic words, and predicates are taken as viewpoint words corresponding to the characteristic words;
when the two words are in a centering relation ATT, taking the center word as a characteristic word and the stationary phrase as a viewpoint word corresponding to the center word;
when the two words are in the relationship ADV in terms of the word, the center word is used as a feature word, and the word is used as a viewpoint word corresponding to the feature word.
4. The method for managing customer experience according to claim 1, wherein S104 specifically comprises:
s1041: extracting word vectors of all feature words through a word embedding technology;
s1042: randomly initializing K clustering centers;
s1043: according to the word vector of each feature word, calculating the distance from the current feature word to the center point of each clustering center:
wherein D is j Representing the distance from the current feature word to the jth clustering center, f i The ith element, c, in the word vector representing the current feature word ij An ith element in the word vector representing the center point of the jth cluster center, and s represents the dimension of the word vector;
s1044: dividing the current feature words into D j In the smallest cluster, and updating the clusterA core;
s1045: and continuously selecting the next feature word until the clustering of all the feature words is completed, and obtaining K product feature clusters.
5. The method for managing customer experience according to claim 1, wherein S106 specifically comprises:
s1061: by comparing the characteristic clusters of each product in pairs and combining a nine-level scale method, a discrimination matrix A is established:
wherein a is ij Indicating the importance level of the ith product feature cluster relative to the jth product feature cluster, a ij The value of (2) can be determined by a nine-pole scale method, and n represents the total number of product characteristic clusters;
s1062: calculating the eigenvector and eigenvalue of the discrimination matrix A:
Aw＝λw→(A-λI)w；
wherein lambda represents the eigenvalue of the discrimination matrix A, w represents the eigenvector of the discrimination matrix A, and the largest eigenvalue is marked as lambda max The eigenvector corresponding to the largest eigenvalue is denoted as w max ，w max ＝(w 1 ,w i ,…,w n )；
S1063: for the feature vector w corresponding to the maximum feature value max Normalization processing:
wherein the normalized vectorAre>Weights respectively representing the characteristic clusters of each product and can be respectively marked as alpha 1 ,α 2 ,…α n 。
6. The method for managing customer experience according to claim 1, wherein S107 specifically comprises:
s1071: counting the number of feature words from each platform contained in each product feature cluster;
s1072: calculating the proximity of comment data of the j-th platform and other platforms:
wherein eta j Representing the proximity of the j-th platform to comment data of other platforms, ρ ij Representing the number of feature words from the jth platform contained in the ith product feature cluster, ρ j Representing the average number of feature words contained in the ith product feature cluster in each platform, and n represents the total number of the product feature clusters;
s1073: the proximity of each platform is standardized to obtain the data credibility of the jth platform:
wherein beta is j The data credibility of the jth platform is represented, and m represents the number of platforms.
7. The method for managing customer experience according to claim 1, wherein S109 is specifically:
calculating the customer experience satisfaction degree of the user for the target product according to the following formula:
wherein S represents customer experience satisfaction, αi represents the weight of the ith product feature cluster, and β j Representing the data trustworthiness of the jth platform,representing the number of forward views from the jth platform contained in the ith product feature cluster, +.>The number of negative viewpoint words from the jth platform contained in the ith product feature cluster is represented, n represents the number of product feature clusters, and m represents the number of platforms.
8. The method for managing customer experience according to claim 1, wherein S110 is specifically:
when the customer experience satisfaction of the user for the target product is lower than the preset satisfaction, an alarm is sent out through at least one of the following modes:
displaying an alarm popup window;
sending an alarm mail to inform related personnel;
or sending an alarm short message to inform related personnel.
9. A customer experience management system for performing the customer experience management method of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311015467.9A CN116739656B (en) | 2023-08-14 | 2023-08-14 | Customer experience management method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311015467.9A CN116739656B (en) | 2023-08-14 | 2023-08-14 | Customer experience management method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116739656A CN116739656A (en) | 2023-09-12 |
| CN116739656B true CN116739656B (en) | 2023-10-20 |
Family
ID=87918994
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311015467.9A Active CN116739656B (en) | 2023-08-14 | 2023-08-14 | Customer experience management method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116739656B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108287922A (en) * | 2018-02-28 | 2018-07-17 | 福州大学 | A kind of text data viewpoint abstract method for digging of fusion topic attribute and emotion information |
| CN111260437A (en) * | 2020-01-14 | 2020-06-09 | 北京邮电大学 | Product recommendation method based on commodity aspect level emotion mining and fuzzy decision |
| CN112328793A (en) * | 2020-11-09 | 2021-02-05 | 北京小米松果电子有限公司 | Comment text data processing method and device and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201118589A (en) * | 2009-06-09 | 2011-06-01 | Ebh Entpr Inc | Methods, apparatus and software for analyzing the content of micro-blog messages |
-
2023
- 2023-08-14 CN CN202311015467.9A patent/CN116739656B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108287922A (en) * | 2018-02-28 | 2018-07-17 | 福州大学 | A kind of text data viewpoint abstract method for digging of fusion topic attribute and emotion information |
| CN111260437A (en) * | 2020-01-14 | 2020-06-09 | 北京邮电大学 | Product recommendation method based on commodity aspect level emotion mining and fuzzy decision |
| CN112328793A (en) * | 2020-11-09 | 2021-02-05 | 北京小米松果电子有限公司 | Comment text data processing method and device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116739656A (en) | 2023-09-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ansari et al. | Analysis of political sentiment orientations on twitter | |
| US20160239763A1 (en) | Motif recognition | |
| US11170061B2 (en) | System for decomposing events from managed infrastructures that includes a reference tool signalizer | |
| CN111125453B (en) | Opinion leader role identification method in social network based on subgraph isomorphism and storage medium | |
| CN106557558B (en) | Data analysis method and device | |
| US10425291B2 (en) | System for decomposing events from managed infrastructures with prediction of a networks topology | |
| Perdana et al. | Combining likes-retweet analysis and naive bayes classifier within twitter for sentiment analysis | |
| US10050910B2 (en) | Application of neural nets to determine the probability of an event being causal | |
| Rosa et al. | Twitter topic fuzzy fingerprints | |
| Alahmadi et al. | Twitter-based recommender system to address cold-start: A genetic algorithm based trust modelling and probabilistic sentiment analysis | |
| US10402428B2 (en) | Event clustering system | |
| CN112182145A (en) | Text similarity determination method, device, equipment and storage medium | |
| Kanev et al. | Sentiment analysis of multilingual texts using machine learning methods | |
| US10979304B2 (en) | Agent technology system with monitoring policy | |
| US10873508B2 (en) | Modularity and similarity graphics system with monitoring policy | |
| CN116739656B (en) | Customer experience management method and system | |
| US11003521B2 (en) | System for decomposing events from managed infrastructures with natural language processing | |
| US10693707B2 (en) | System for decomposing events from managed infrastructures with semantic clustering | |
| CN114445043B (en) | Open ecological cloud ERP-based heterogeneous graph user demand accurate discovery method and system | |
| Bijoy et al. | Classifying Gender Based on Life Partner Choosing Factor using Supervised Machine Learning | |
| US10853820B2 (en) | Method and apparatus for recommending topic-cohesive and interactive implicit communities in social customer relationship management | |
| Dinh et al. | A framework to discover potential ideas of new product development from crowdsourcing application | |
| US10715379B2 (en) | System for decomposing events from managed infrastructures with bitwise operation | |
| Dey et al. | Comparative Exploration Of Prediction Algorithms For Sentiment Analysis Using NLP | |
| Vásquez et al. | Using centrality measures to improve the classification performance of tweets during natural disasters |
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 |