CN115309864A - Intelligent sentiment classification method and device for comment text, electronic equipment and medium - Google Patents
Intelligent sentiment classification method and device for comment text, electronic equipment and medium Download PDFInfo
- Publication number
- CN115309864A CN115309864A CN202210961855.5A CN202210961855A CN115309864A CN 115309864 A CN115309864 A CN 115309864A CN 202210961855 A CN202210961855 A CN 202210961855A CN 115309864 A CN115309864 A CN 115309864A
- Authority
- CN
- China
- Prior art keywords
- text
- vector
- word vector
- comment
- word
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- 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
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention relates to the field of intelligent decision making, and discloses an intelligent sentiment classification method for comment texts, which comprises the following steps: analyzing the evaluation object category of the comment text, and extracting comment dimension information of the evaluation object category; performing word segmentation processing on the comment text to obtain text word segmentation; converting the comment dimension information and the text segmentation into a dimension word vector and a text word vector; calculating the attention score of the text word vector, and fusing the attention score and the text word vector to obtain a text fused word vector; calculating a vector inner product matrix of the text fusion word vector and the dimension word vector, and marking the vector weight of an inner product vector in the vector inner product matrix; re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and performing pooling treatment on the target fused word vector to obtain a pooled fused word vector; and calculating the text category probability of the pooling fusion word vector to determine the text emotion category of the comment text. The method and the device can improve the emotion classification accuracy of the comment text.
Description
Technical Field
The invention relates to the field of intelligent decision, in particular to an emotion intelligent classification method and device for comment texts, electronic equipment and a medium.
Background
With the development of network technology, more and more internet users share experience and make comments on various websites, and the comment texts express the opinions and emotions of commentators and imply huge commercial values. Various shopping websites, forums, blogs, microblogs and the like provide a wide platform for expressing opinions and exchanging information for internet users, people gradually get used to share experiences and make comments on various websites, and directly express positive or negative, supported or objected emotions of people, such as book comment, movie comment, evaluation on a hotel or use experience of a certain mobile phone. Users are also used to get information from various comments on the internet and find reference opinions for certain decisions of the users.
Currently, in the industry, text comments are often classified based on a chapter-level or sentence-level sentiment classification, the chapter-level is a total sentiment category calculated for the whole text comments, the sentence-level is a sentiment category classification for each sentence in the whole text comments, but the text sentiment classification realized through the chapter-level or sentence-level cannot show a specific sentiment direction in a text from a finer-grained angle, for example, a sentence may contain multiple dimensions, each dimension has its own sentiment category, and if the text sentiment classification is classified from the chapter-level or sentence-level angle, the sentiment category of the text comments cannot be accurately determined, which results in inaccurate text sentiment classification.
Disclosure of Invention
The invention provides an intelligent sentiment classification method, device, electronic equipment and medium for comment texts, and mainly aims to improve the sentiment classification accuracy of the comment texts.
In order to achieve the above object, the invention provides an intelligent sentiment classification method for comment texts, which comprises the following steps:
obtaining a comment text, analyzing an evaluation object category of the comment text, and extracting comment dimension information corresponding to the evaluation object category;
performing word segmentation processing on the comment text by using a word segmentation tool to obtain text word segmentation;
converting the comment dimension information and the text segmentation into a dimension word vector and a text word vector respectively;
calculating an attention score of the text word vector by using an attention mechanism in a trained text emotion classification model, and fusing the attention score and the text word vector to obtain a text fusion word vector;
calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in a trained text emotion classification model, and marking the vector weight of each inner product vector in the vector inner product matrix;
re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and performing pooling treatment on the target fused word vector by using a pooling layer in the trained text emotion classification model to obtain a pooled fused word vector;
and calculating the text emotion category probability of the pooling fusion word vector by using a multilayer perceptron in a trained text emotion classification model, and determining the emotion category of the comment text by using an output layer in the trained text emotion classification model according to the text emotion category probability.
Optionally, the analyzing the evaluation object category of the comment text includes:
extracting original characteristic words of the comment text;
extracting feature object words in the original feature words, and converting the feature object words into feature category words;
screening out classification characteristic words meeting preset conditions from the original characteristic words;
and determining the evaluation object type of the comment text according to the characteristic type words and the classification characteristic words.
Optionally, the calculating an attention score of the text word vector by using an attention mechanism in the trained text emotion classification model includes:
creating a dimension vector corresponding to the text word vector by using a vector matrix in the attention mechanism;
and calculating the vector weight of the dimension vector by using a dot product function in the attention mechanism, and taking the vector weight as the attention score of the text word vector.
Optionally, the fusing the attention score with the text word vector to obtain a text fused word vector includes:
normalizing the attention score by using a preset normalization function to obtain a weight coefficient, and creating a numerical vector corresponding to the text word vector;
and multiplying the weight coefficient by the numerical value vector to obtain a text fusion word vector.
Optionally, the preset normalization function includes:
wherein, y k Represents a weight coefficient, a k Attention score, a, representing the k-th text word vector i Denotes the attention score of the ith text word vector, n denotes the number of attention scores, and e denotes an infinite acyclic decimal.
Optionally, the calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in the trained text emotion classification model includes:
creating a text fusion matrix of the text fusion word vector by using a vector matrix conversion method in the conversion layer, and creating a dimension word matrix of the dimension word vector by using the vector matrix conversion method;
and performing inner product calculation on the text fusion matrix and the dimension word matrix by using an inner product function in the conversion layer to obtain the vector inner product matrix.
Optionally, the determining, according to the text category probability, the text emotion category of the comment text by using the output layer in the trained text emotion classification model includes:
and determining the text category of the comment text by utilizing a feed-forward neural network in the output layer according to the text category probability, and determining the text emotion category of the comment text by utilizing a preset emotion tag threshold mapping relation table based on the text category.
In order to solve the above problem, the present invention further provides an apparatus for intelligently classifying sentiment of comment text, the apparatus comprising:
the category identification module is used for acquiring a comment text, analyzing the evaluation object category of the comment text and extracting comment dimension information corresponding to the evaluation object category;
the text word segmentation module is used for performing word segmentation processing on the comment text by using a word segmentation tool to obtain text word segmentation;
the word vector conversion module is used for converting the comment dimension information and the text segmentation into a dimension word vector and a text word vector respectively;
the word vector fusion module is used for calculating the attention score of the text word vector by using an attention mechanism in the trained text emotion classification model and fusing the attention score and the text word vector to obtain a text fusion word vector;
the vector inner product module is used for calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by utilizing a conversion layer in the trained text emotion classification model, and marking the vector weight of each inner product vector in the vector inner product matrix;
the pooling processing module is used for re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and pooling the target fused word vector by using a pooling layer in the trained text sentiment classification model to obtain a pooled fused word vector;
and the emotion category judging module is used for calculating the text emotion category probability of the pooling fusion word vector by using a multi-layer perceptron in the trained text emotion classification model, and determining the emotion category of the comment text by using an output layer in the trained text emotion classification model according to the text emotion category probability.
In order to solve the above problem, the present invention also provides an electronic device, including:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to implement the method for sentiment intelligent classification of comment text described above.
In order to solve the above problem, the present invention further provides a computer-readable storage medium, in which at least one computer program is stored, and the at least one computer program is executed by a processor in an electronic device to implement the above-mentioned method for sentiment intelligent classification of comment text.
The method and the device for identifying the comment text provide an operation object for the subsequent method implementation by identifying the evaluation object type of the obtained comment text, and obtain comment data of the comment object type from multiple dimensional directions according to comment dimensional information corresponding to the evaluation object type, so that the fine-grained information splitting of the comment text corresponding to the subsequent evaluation object type is guaranteed. Utilizing a word segmentation tool to perform word segmentation processing on the comment text to obtain text word segmentation, and respectively converting the comment dimension information and the text word segmentation into a dimension word vector and a text word vector, wherein the dimension word vector and the text word vector are used for preprocessing the comment text to provide input for a subsequent text emotion classification model and further processing the input as a guarantee; in the second embodiment of the present invention, an attention mechanism in a trained text sentiment classification model is used to calculate an attention score of the text word vector, which can be used to describe an attention degree of each text participle in the comment text, the attention score and the text word vector are fused to generate a text fusion word vector that more truly reflects an actual application scene, an input support can be provided for a conversion layer in a subsequent text sentiment classification model, a vector inner product matrix of the text fusion word vector and the dimension word vector is calculated by using the conversion layer in the trained text sentiment classification model to provide a support for subsequently obtaining a weight coefficient of the text fusion word vector in the aspect of dimension words, a target fusion word vector is subsequently generated, a vector weight of each inner product vector in the vector inner product matrix is marked to determine a weight of the text fusion word vector in the aspect of dimension words, and the comment text are subsequently associated to more fully express the comment text, so as to fine-grain the comment text; further, in the embodiment of the present invention, the vector weight and the text fusion word vector are re-fused to obtain a target fusion word vector, so that finer-grained expression of the comment text can be obtained, accuracy of subsequent text emotion classification is improved, a pooling layer in a trained text emotion classification model is used to perform pooling processing on the target fusion word vector, and the obtained pooling fusion word vector can be used as an input of an activation function of a multilayer perceptron in the subsequent text emotion classification model to perform further processing. The text category probability of the pooling fusion word vector is calculated by using a multilayer perceptron in the trained text emotion classification model, and the text emotion category of the comment text is determined by using an output layer in the trained text emotion classification model according to the text category probability, so that fine-grained evaluation of the comment text can be realized, the accuracy of subsequent text emotion classification is improved, and more accurate reference opinions are provided for user decision making. Therefore, the sentiment intelligent classification method, the sentiment intelligent classification device, the electronic equipment and the storage medium for the comment text provided by the embodiment of the invention can improve the sentiment classification accuracy of the comment text.
Drawings
Fig. 1 is a schematic flowchart of an intelligent sentiment classification method for comment texts according to an embodiment of the present invention;
fig. 2 is a schematic block diagram of an apparatus for intelligently classifying emotion in comment text according to an embodiment of the present invention;
fig. 3 is a schematic diagram of an internal structure of an electronic device implementing the intelligent sentiment classification method for comment texts according to an embodiment of the present invention;
the implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The embodiment of the invention provides an intelligent sentiment classification method for comment texts. The execution subject of the sentiment intelligent classification method for comment texts includes but is not limited to at least one of a server, a terminal and other electronic devices which can be configured to execute the method provided by the embodiment of the invention. In other words, the emotion intelligent classification method for comment texts may be executed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like. The server may be an independent server, or may be a cloud server that provides basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a Content Delivery Network (CDN), a big data and artificial intelligence platform, and the like.
Fig. 1 is a schematic flow chart of an emotion intelligent classification method for comment texts according to an embodiment of the present invention. In the embodiment of the invention, the intelligent sentiment classification method for comment texts comprises the following steps of S1-S7:
s1, obtaining a comment text, analyzing an evaluation object type of the comment text, and extracting comment dimension information corresponding to the evaluation object type;
in the embodiment of the invention, the comment text is obtained to provide an operation object for the implementation of a subsequent method. The comment text refers to the expression form of the written language to be evaluated, such as sentences, paragraphs, chapters and the like.
Further, the comment text may be obtained by a crawler technology according to an optional embodiment of the present invention. The crawler technology refers to a web crawler, which is a program or script for automatically capturing web information according to a certain rule.
In the embodiment of the invention, a basis can be provided for subsequently determining the dimension information of the comment text by analyzing the evaluation object category of the comment text. The evaluation object category refers to a category to which an object or an attribute of the object to which the comment text is directed belongs.
Further, in an optional embodiment of the present invention, the analyzing the evaluation object category of the comment text includes: extracting original characteristic words of the comment text; extracting feature object words in the original feature words, and converting the feature object words into feature category words; screening out classification characteristic words meeting preset conditions from the original characteristic words; and determining the evaluation object category of the comment text by combining the characteristic category words and the classification characteristic words.
The original feature words refer to feature words extracted from the text at first. The feature object word refers to a noun representing an object in the original feature word. The classification characteristic words are words used for describing the attributes with classification information, such as dish names, good eating, environment, silence, price, high price and the like.
Further, in an optional embodiment of the present invention, the original feature words of the comment text may be implemented by a preset text analysis algorithm, where the preset text analysis algorithm includes an Information Gain (IG) and a chi-square check (chi-square) algorithm, and is configured to represent a text and select feature items thereof, so as to quantify feature words extracted from the text to represent text information, and the feature category words of the comment text may be implemented by a preset mapping conversion method, where the preset mapping conversion method includes an object word-category word mapping relationship, and is used to describe a method of a correspondence relationship between elements of two sets of elements that are transformed with each other, such as a method of mapping price, environment, location, and service of object words in the comment text to a category word service industry. The classification feature words of the comment text can be realized through a preset word class mapping relation of a corpus, such as mapping precious or cheap words into price classes, mapping far or near words into distance classes, mapping prices and distances into service industry classes, and the like. The corpus is a large-scale electronic text library which is scientifically sampled and processed, and stored therein are language materials which actually appear in the practical use of languages.
Further, in an optional embodiment of the present invention, the evaluation object category of the comment text is obtained by performing semantic overlapping on the feature category word and the classification feature word.
In the embodiment of the invention, comment data of the comment object category can be acquired from multiple dimensional directions by extracting comment dimensional information corresponding to the evaluation object category, and the fine-grained information splitting of comment texts corresponding to the subsequent evaluation object category is ensured. The comment dimension information refers to a relative or parallel information description associated with the object category at a certain level, for example, the comment dimension information about the evaluation text of a restaurant is dishes, price, environment, location, service, and the like.
It should be understood that, in an implementation of the present invention, the extracting comment dimension information corresponding to the evaluation object category includes: and matching the evaluation object category with the dimension category in a pre-constructed category-dimension relation table, and taking the evaluation dimension corresponding to the successfully matched dimension category as comment dimension information corresponding to the evaluation object category.
The pre-constructed category-dimension relation table is a data table obtained by performing relation mapping on a historical object category and a historical evaluation dimension, the historical object category and the historical evaluation dimension can be acquired through a big data technology, the historical object category and the historical evaluation dimension can be mapped through a relation mapping algorithm, and the relation mapping algorithm can be compiled through a python language.
Further, in an optional embodiment of the present invention, matching between the evaluation object class and a dimension class in a pre-constructed class-dimension relationship table may be implemented by a matching algorithm, such as a cosine similarity matching algorithm.
S2, performing word segmentation processing on the comment text by using a word segmentation tool to obtain text word segmentation;
in the embodiment of the invention, the word segmentation is carried out on the comment text by using the word segmentation tool, and the obtained text word segmentation is the preprocessing of the comment text, so that the guarantee is provided for further operation of subsequently obtaining the word vector of the comment text.
The word segmentation tool is a tool for recombining continuous word sequences into word sequences according to a certain standard, such as a jieba word segmentation tool, an ltp word segmentation tool, an ir word segmentation tool and the like. The text word segmentation means that a Chinese character sequence is segmented into a single word. For example, the text segment of "the restaurant is good at or too expensive" is that the restaurant is good at or too expensive.
S3, converting the comment dimension information and the text participles into dimension word vectors and text word vectors respectively;
in the embodiment of the invention, the comment dimension information and the text participles are respectively converted into the dimension word vector and the text word vector to be used as the input of a subsequent text emotion classification model, so that a basis is provided for further processing. The text word vector is a vector that maps words or phrases in a natural language vocabulary to a real number space. Conceptually, it involves mathematical embedding from a one-dimensional space to a multi-dimensional continuous vector space of each word. The comment dimension information is 'linked' with the object, and is beneficial to better understanding the information of the object and the form content thereof, such as 'dishes', 'prices', and 'services', and the like, and is comment dimension information of 'restaurants'.
Further, in an optional embodiment of the present invention, the converting the comment dimension information and the text participle into a dimension word vector and a text word vector, respectively, may be implemented by calculating word vectors of the comment dimension information and the text participle by using a trained vector conversion model, such as Skip-gram, CBOW, LBL, NNLM, C & W, and Glove.
S4, calculating the attention score of the text word vector by using an attention mechanism in the trained text emotion classification model, and fusing the attention score and the text word vector to obtain a text fusion word vector;
in the embodiment of the invention, the attention score of the text word vector is calculated by using the attention mechanism in the trained text emotion classification model, so that the attention degree of each text participle in the comment text can be described, and support is provided for subsequently generating the text fusion word vector which more truly reflects the actual application scene. The text emotion classification model refers to a formal expression mode obtained by performing abstract training classification on the features and rules of a text, such as a Support Vector Machine (SVM), a K Nearest Neighbor (KNN) model and other text emotion classification models. The attention mechanism stems from research on human vision, in which a human being selectively focuses on a portion of all information while ignoring other visible information due to a bottleneck in information processing, and is generally referred to as the attention mechanism.
Further, in an optional embodiment of the present invention, the calculating an attention score of the text word vector by using an attention mechanism in the trained text emotion classification model includes: creating a dimension vector corresponding to the text word vector by using a vector matrix in the attention mechanism; and calculating the vector weight of the dimension vector by using a dot product function in the attention mechanism, and taking the vector weight as the attention score of the text word vector.
The dot product function refers to a binary operation that accepts two vectors on a real number R and returns a real-valued scalar, and is implemented by multiplying and summing corresponding components of the two vectors of the same dimension.
In the embodiment of the invention, the text fusion word vector obtained by fusing the attention score and the text word vector can provide input support for a conversion layer in the subsequent text emotion classification model.
Further, in an optional embodiment of the present invention, the fusing the attention score and the text word vector to obtain a text fused word vector includes: normalizing the attention score by using a preset normalization function to obtain a weight coefficient, and creating a numerical vector corresponding to the text word vector; and multiplying the weight coefficient by the numerical value vector to obtain a text fusion word vector.
The weighting coefficient is used for indicating the importance degree of a certain text word vector item in a text word vector system, can be used for displaying the importance degrees of a plurality of text word vectors in the total amount of the text word vectors, and is respectively given different proportionality coefficients.
Further, in an optional embodiment of the present invention, the preset normalization function includes:
wherein, y k Represents a weight coefficient, a k Attention score, a, representing the k-th text word vector i An attention score representing the ith text word vector, n representing the number of attention scores, and e representing an infinite acyclic decimal.
Further, in an optional embodiment of the present invention, a principle of creating a numerical vector corresponding to the text word vector is the same as that of creating the query vector and the identifier vector, and further details are not described herein.
S5, calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in the trained text emotion classification model, and marking the vector weight of each inner product vector in the vector inner product matrix;
in the embodiment of the invention, by utilizing the conversion layer in the trained text emotion classification model, the vector inner product matrix of the text fusion word vector and the dimension word vector is calculated, so that support can be provided for subsequently obtaining the weight coefficient of the text fusion word vector in the aspect of dimension words, and a target fusion word vector is subsequently generated. The conversion layer is optionally a processing layer for performing further feature extraction and the like on the output of the SA layer by using the FFN layer of the transform model.
Further, in the embodiment of the present invention, the calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in the trained text emotion classification model includes: creating a text fusion matrix of the text fusion word vector by using a vector matrix conversion method in the conversion layer, and creating a dimension word matrix of the dimension word vector by using the vector matrix conversion method; and performing inner product calculation on the text fusion matrix and the dimension word matrix by using an inner product function in the conversion layer to obtain the vector inner product matrix.
The vector matrix conversion method is a method for converting a group of vectors into a matrix, and can be implemented by arranging the vectors in sequence as a row or a column of the matrix according to the same rule.
In the embodiment of the invention, the weight of the text fusion word vector in the aspect of the dimension word can be determined by marking the vector weight of each inner product vector in the vector inner product matrix, so that the comment text is associated with the dimension in the follow-up process to be more fully expressed, and the comment text is commented more finely.
Further, in an optional embodiment of the present invention, the marking the vector weight of each inner product vector in the vector inner product matrix may be implemented by determining a matrix element position corresponding to each inner product vector in the vector inner product matrix according to a matrix multiplication principle.
S6, re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and performing pooling treatment on the target fused word vector by using a pooling layer in the trained text emotion classification model to obtain a pooled fused word vector;
in the embodiment of the invention, the vector weight and the text fusion word vector are re-fused to obtain the target fusion word vector, so that the comment text can be expressed in a finer granularity, and the accuracy of subsequent text emotion classification is improved.
Further, in an optional embodiment of the present invention, the re-fusing the vector weight and the text fused word vector to obtain the target fused word vector may be implemented by performing dot product calculation on the vector weight and the text fused word vector.
In the embodiment of the invention, the target fusion word vector is subjected to pooling treatment by utilizing the pooling layer in the trained text emotion classification model, and the obtained pooling fusion word vector can be used as the input of the activation function of the multilayer perceptron in the subsequent text emotion classification model for further treatment. Wherein the pooling layer is obtained by sub-sampling the data to down-sample a large matrix into a small matrix to reduce the amount of computation and prevent over-fitting. The pooling is a series of operations performed on input data by a pooling layer, and includes pooling for maximum, mean, random, median, and combined.
And S7, calculating the text category probability of the pooling fusion word vector by using a multi-layer perceptron in the trained text emotion classification model, and determining the text emotion category of the comment text by using an output layer in the trained text emotion classification model according to the text category probability.
In the embodiment of the invention, the text category probability of the pooling fusion word vector is calculated by utilizing the multi-layer perceptron in the trained text emotion classification model, so that guarantee can be provided for subsequently determining the text emotion category of the comment text. The multilayer perceptron is a neural network which comprises at least one hidden layer and is composed of fully-connected layers, and the output of each hidden layer is transformed through an activation function. The activation function refers to a nonlinear function running on a hidden layer neuron of the artificial neural network and is responsible for mapping the input of the neuron to the output, such as a sigmoid function, a ReLU function, a tanh function and other activation functions.
Further, in an optional embodiment of the present invention, the text category probability of the pooled fusion word vector is calculated by using the following formula:
where x is a neuron input in the artificial neural network, f (x) is a neuron output that functionally maps the neuron input x, and R represents a real number domain.
In the embodiment of the invention, the text emotion type of the comment text is determined by utilizing the output layer in the trained text emotion classification model according to the text type probability, so that the fine-grained evaluation of the comment text can be realized, and a more accurate reference opinion is provided for the decision of a user.
Further, in an optional embodiment of the present invention, the determining, according to the text category probability and by using an output layer in the trained text emotion classification model, a text emotion category of the comment text includes: and determining the text category of the comment text by utilizing a feed-forward neural network in the output layer according to the text category probability, and determining the text emotion category of the comment text by utilizing a preset emotion label threshold mapping relation table based on the text category.
The preset emotion label threshold mapping relation table can be constructed according to service scenes, for example, emotion labels are set to be of four types: positive, neutral, negative, not relevant, the emotion tag threshold mapping relationship table may be as follows:
f:A→B,f 1 :A 1 →B 1 ,f 2 :A 2 →B 2 ,f 3 :A 3 →B 3 ,f 4 :A 4 →B 4 ；
wherein, a = { a = 1, A 2, A 3 ,A 4 }，B＝{B 1, B 2, B 3 ,B 4 }，
A 2 ＝{x i ≥T h ,0.5≤nin<0.75，A3＝xi≥Th,0.25≤nin<0.5，A4＝xi≥Th,0≤nin<0.25，x＝x1,x2,x3…,xn，0≤x i≤ 1,i＝1,2,3…,n；B 1 = active, B 2 = neutral, B 3 = passive, B 4 = uncorrelated.
Wherein f represents an emotion label threshold value mapping relation table, x represents the text dimension probability (is a probability vector) of a text word vector to be commented, and x i Representing a text dimension probability component, T, corresponding to the ith comment dimension h Is the tag threshold, optionally T h =0.5, and can also be set according to practical application scenarios, where n is the dimension of the vector of the word in the paper to be evaluated, and n is the dimension of the vector i And expressing the number of vector components meeting the threshold condition in the word vector of the paper to be evaluated corresponding to the ith comment dimension.
The method and the device for identifying the comment text provide an operation object for the subsequent method implementation by identifying the evaluation object type of the obtained comment text, and obtain comment data of the comment object type from multiple dimensional directions according to comment dimensional information corresponding to the evaluation object type, so that the fine-grained information splitting of the comment text corresponding to the subsequent evaluation object type is guaranteed. Utilizing a word segmentation tool to perform word segmentation processing on the comment text to obtain text word segmentation, and respectively converting the comment dimension information and the text word segmentation into a dimension word vector and a text word vector, wherein the dimension word vector and the text word vector are used for preprocessing the comment text to provide input for a subsequent text emotion classification model and further processing the input as a guarantee; in the embodiment of the invention, the attention score of the text word vector is calculated by using an attention mechanism in a trained text sentiment classification model, which can be used for describing the attention degree of each text participle in the comment text, the attention score and the text word vector are fused to generate a text fusion word vector which more truly reflects the actual application scene, so that input support can be provided for a conversion layer in the subsequent text sentiment classification model, a vector inner product matrix of the text fusion word vector and the dimension word vector is calculated by using the conversion layer in the trained text sentiment classification model, so that support can be provided for subsequently obtaining a weight coefficient of the text fusion word vector in the aspect of dimension words, a target fusion word vector is subsequently generated, the vector weight of each inner product vector in the vector inner product matrix is marked, so that the weight of the text fusion word vector in the aspect of dimension words can be determined, the comment text is subsequently associated with the dimension in the comment text to more fully express the comment text, and the comment text is more fine-grained; further, in the embodiment of the present invention, the vector weight and the text fusion word vector are re-fused to obtain a target fusion word vector, so that finer-grained expression of the comment text can be obtained, accuracy of subsequent text emotion classification is improved, a pooling layer in a trained text emotion classification model is used to perform pooling processing on the target fusion word vector, and the obtained pooling fusion word vector can be used as an input of an activation function of a multilayer perceptron in the subsequent text emotion classification model to perform further processing. The text category probability of the pooled fusion word vector is calculated by using a multilayer perceptron in the trained text sentiment classification model, and the text sentiment category of the comment text is determined by using an output layer in the trained text sentiment classification model according to the text category probability, so that the fine-grained evaluation of the comment text can be realized, the accuracy of subsequent text sentiment classification is improved, and more accurate reference sentiments are provided for user decision making. Therefore, the sentiment intelligent classification method, the sentiment intelligent classification device, the electronic equipment and the storage medium for the comment text provided by the embodiment of the invention can improve the sentiment classification accuracy of the comment text.
Fig. 2 is a functional block diagram of the apparatus for classifying comment texts according to the present invention.
The intelligent sentiment classification device 100 for comment texts can be installed in electronic equipment. According to the realized functions, the intelligent sentiment classification device for comment texts can comprise a category identification module 101, a text word segmentation module 102, a word vector conversion module 103, a word vector fusion module 104, a vector inner product module 105, a pooling processing module 106 and a sentiment category judgment module 107. The module of the present invention, which may also be referred to as a unit, refers to a series of computer program segments that can be executed by a processor of an electronic device and can perform a fixed function, and is stored in a memory of the electronic device.
In the present embodiment, the functions regarding the respective modules/units are as follows:
the category identification module 101 is configured to acquire a comment text, analyze an evaluation object category of the comment text, and extract comment dimension information corresponding to the evaluation object category;
the text word segmentation module 102 is configured to perform word segmentation processing on the comment text by using a word segmentation tool to obtain text words;
the word vector conversion module 103 is configured to convert the comment dimension information and the text segmentation into a dimension word vector and a text word vector, respectively;
the word vector fusion module 104 is configured to calculate an attention score of the text word vector by using an attention mechanism in the trained text emotion classification model, and fuse the attention score and the text word vector to obtain a text fusion word vector;
the vector inner product module 105 is configured to calculate a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in a trained text emotion classification model, and mark a vector weight of each inner product vector in the vector inner product matrix;
the pooling processing module 106 is configured to perform re-fusion on the vector weights and the text fused word vectors to obtain target fused word vectors, and perform pooling processing on the target fused word vectors by using a pooling layer in the trained text emotion classification model to obtain pooled fused word vectors;
the emotion classification judging module 107 is configured to calculate a text emotion classification probability of the pooled fusion word vector by using a multi-layer perceptron in a trained text emotion classification model, and determine an emotion classification of the comment text by using an output layer in the trained text emotion classification model according to the text emotion classification probability.
In detail, when the modules in the apparatus 100 for intelligently classifying sentiment texts according to the embodiment of the present invention are used, the same technical means as the above-mentioned method for intelligently classifying sentiment texts in fig. 1 are adopted, and the same technical effect can be produced, which is not described herein again.
As shown in fig. 3, the structural diagram of the electronic device 1 implementing the emotion intelligent classification method for comment texts is shown.
The electronic device 1 may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13, and may further include a computer program, such as an emotion intelligent classification program for comment text, stored in the memory 11 and executable on the processor 10.
In some embodiments, the processor 10 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or may be composed of a plurality of integrated circuits packaged with the same function or different functions, and includes one or more Central Processing Units (CPUs), a microprocessor, a digital Processing chip, a graphics processor, a combination of various control chips, and the like. The processor 10 is a Control Unit (Control Unit) of the electronic device 1, connects various components of the electronic device 1 by using various interfaces and lines, and executes various functions and processes data of the electronic device 1 by running or executing programs or modules (for example, an emotion intelligence classification program for executing comment text, etc.) stored in the memory 11 and calling data stored in the memory 11.
The memory 11 includes at least one type of readable storage medium including flash memory, removable hard disks, multimedia cards, card-type memory (e.g., SD or DX memory, etc.), magnetic memory, magnetic disks, optical disks, etc. The memory 11 may in some embodiments be an internal storage unit of the electronic device 1, such as a removable hard disk of the electronic device 1. The memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in mobile hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, which are provided on the electronic device 1. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device 1. The memory 11 may be used not only to store application software installed in the electronic device 1 and various types of data, such as codes of an emotion intelligent classification program for commenting on texts, but also to temporarily store data that has been output or is to be output.
The communication bus 12 may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is arranged to enable connection communication between the memory 11 and at least one processor 10 or the like.
The communication interface 13 is used for communication between the electronic device 1 and other devices, and includes a network interface and an employee interface. Optionally, the network interface may include a wired interface and/or a wireless interface (e.g., WI-FI interface, bluetooth interface, etc.), which are generally used for establishing a communication connection between the electronic device 1 and other electronic devices 1. The employee interface may be a Display (Display), an input unit, such as a Keyboard (Keyboard), and optionally a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, or the like. The display, which may also be referred to as a display screen or display unit, is suitable for displaying information processed in the electronic device 1 and for displaying a visual staff interface.
Fig. 3 only shows the electronic device 1 with components, and it will be understood by those skilled in the art that the structure shown in fig. 3 does not constitute a limitation of the electronic device 1, and may comprise fewer or more components than shown, or some components may be combined, or a different arrangement of components.
For example, although not shown, the electronic device 1 may further include a power supply (such as a battery) for supplying power to each component, and preferably, the power supply may be logically connected to the at least one processor 10 through a power management device, so as to implement functions of charge management, discharge management, power consumption management, and the like through the power management device. The power supply may also include any component of one or more dc or ac power sources, recharging devices, power failure detection circuitry, power converters or inverters, power status indicators, and the like. The electronic device 1 may further include various sensors, a bluetooth module, a Wi-Fi module, and the like, which are not described herein again.
It is to be understood that the embodiments described are for illustrative purposes only and that the scope of the claimed invention is not limited to this configuration.
The emotion intelligent classification program of the comment text stored in the memory 11 in the electronic device 1 is a combination of a plurality of computer programs, and when running in the processor 10, can realize that:
obtaining a comment text, analyzing an evaluation object category of the comment text, and extracting comment dimension information corresponding to the evaluation object category;
performing word segmentation processing on the comment text by using a word segmentation tool to obtain text word segmentation;
converting the comment dimension information and the text segmentation into a dimension word vector and a text word vector respectively;
calculating the attention score of the text word vector by using an attention mechanism in a trained text emotion classification model, and fusing the attention score and the text word vector to obtain a text fused word vector;
calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in a trained text sentiment classification model, and marking the vector weight of each inner product vector in the vector inner product matrix;
re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and performing pooling treatment on the target fused word vector by using a pooling layer in the trained text emotion classification model to obtain a pooled fused word vector;
and calculating the text emotion category probability of the pooling fusion word vector by using a multilayer perceptron in a trained text emotion classification model, and determining the emotion category of the comment text by using an output layer in the trained text emotion classification model according to the text emotion category probability.
Specifically, the processor 10 may refer to the description of the relevant steps in the embodiment corresponding to fig. 1 for a specific implementation method of the computer program, which is not described herein again.
Further, the integrated modules/units of the electronic device 1, if implemented in the form of software functional units and sold or used as separate products, may be stored in a non-volatile computer-readable storage medium. The computer readable storage medium may be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying said computer program code, recording medium, U-disk, removable hard disk, magnetic disk, optical disk, computer Memory, read-Only Memory (ROM).
The invention also provides a computer-readable storage medium, in which a computer program is stored, which computer program, when executed by a processor of an electronic device 1, enables:
obtaining a comment text, analyzing an evaluation object category of the comment text, and extracting comment dimension information corresponding to the evaluation object category;
performing word segmentation processing on the comment text by using a word segmentation tool to obtain text word segmentation;
converting the comment dimension information and the text segmentation into a dimension word vector and a text word vector respectively;
calculating the attention score of the text word vector by using an attention mechanism in a trained text emotion classification model, and fusing the attention score and the text word vector to obtain a text fused word vector;
calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in a trained text emotion classification model, and marking the vector weight of each inner product vector in the vector inner product matrix;
re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and performing pooling treatment on the target fused word vector by using a pooling layer in the trained text emotion classification model to obtain a pooled fused word vector;
and calculating the text emotion category probability of the pooling fusion word vector by using a multilayer perceptron in a trained text emotion classification model, and determining the emotion category of the comment text by using an output layer in the trained text emotion classification model according to the text emotion category probability.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is only one logical functional division, and other divisions may be realized in practice.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional module.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof.
The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.
The embodiment of the invention can acquire and process related data based on an artificial intelligence technology. Among them, artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.
Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. A plurality of units or means recited in the system claims may also be implemented by one unit or means in software or hardware. The terms second, etc. are used to denote names, but not any particular order.
Finally, it should be noted that the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims (10)
1. An emotion intelligent classification method for comment texts, which is characterized by comprising the following steps:
obtaining a comment text, analyzing an evaluation object category of the comment text, and extracting comment dimension information corresponding to the evaluation object category;
performing word segmentation processing on the comment text by using a preset word segmentation tool to obtain text word segmentation;
converting the comment dimension information and the text segmentation into a dimension word vector and a text word vector respectively;
calculating an attention score of the text word vector by using an attention mechanism in a trained text emotion classification model, and fusing the attention score and the text word vector to obtain a text fusion word vector;
calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in a trained text emotion classification model, and marking the vector weight of each inner product vector in the vector inner product matrix;
re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and performing pooling treatment on the target fused word vector by using a pooling layer in the trained text emotion classification model to obtain a pooled fused word vector;
and calculating the text category probability of the pooling fusion word vector by using a multilayer perceptron in the trained text emotion classification model, and determining the text emotion category of the comment text by using an output layer in the trained text emotion classification model according to the text category probability.
2. The method for intelligently classifying emotion of comment text according to claim 1, wherein said analyzing evaluation object categories of said comment text includes:
extracting original characteristic words of the comment text;
extracting feature object words in the original feature words, and converting the feature object words into feature category words;
screening out classification characteristic words meeting preset conditions from the original characteristic words;
and determining the evaluation object type of the comment text according to the characteristic type words and the classification characteristic words.
3. The method for intelligent classification of emotion in comment text as recited in claim 1, wherein said calculating an attention score for said text word vector using an attention mechanism in a trained text emotion classification model comprises:
creating a dimension vector corresponding to the text word vector by using a vector matrix in the attention mechanism;
and calculating the vector weight of the dimension vector by using a dot product function in the attention mechanism, and taking the vector weight as the attention score of the text word vector.
4. The method for intelligently classifying emotion of comment text as claimed in claim 1, wherein said fusing said attention score with said text word vector to obtain a text fused word vector comprises:
normalizing the attention score by using a preset normalization function to obtain a weight coefficient, and creating a numerical vector corresponding to the text word vector;
and multiplying the weight coefficient by the numerical value vector to obtain a text fusion word vector.
5. The method for intelligently classifying emotion of comment text as recited in claim 1, wherein said preset normalization function comprises:
wherein, y k Represents a weight coefficient, a k Attention score, a, representing the k-th text word vector i Represents the ithThe attention score of the text word vector, n represents the number of attention scores, and e represents an infinite acyclic decimal.
6. The method for intelligent sentiment classification of comment text according to claim 1, wherein the calculating of the vector inner product matrix of the text fusion word vector and the dimension word vector by using a conversion layer in a trained text sentiment classification model comprises:
creating a text fusion matrix of the text fusion word vector by using a vector matrix conversion method in the conversion layer, and creating a dimension word matrix of the dimension word vector by using the vector matrix conversion method;
and performing inner product calculation on the text fusion matrix and the dimension word matrix by using an inner product function in the conversion layer to obtain the vector inner product matrix.
7. The method for intelligent sentiment classification of comment texts according to claim 1, wherein the determining the text sentiment classification of the comment texts by using an output layer in the trained text sentiment classification model according to the text category probability comprises:
determining the text category of the comment text by utilizing a feed-forward neural network in the output layer according to the text category probability;
and determining the text emotion category of the comment text by utilizing a preset emotion tag threshold mapping relation table based on the text category.
8. An apparatus for intelligent sentiment classification of comment text, the apparatus comprising:
the category identification module is used for acquiring the comment text, analyzing the evaluation object category of the comment text and extracting comment dimension information corresponding to the evaluation object category;
the text word segmentation module is used for performing word segmentation processing on the comment text by using a word segmentation tool to obtain text word segmentation;
the word vector conversion module is used for respectively converting the comment dimension information and the text segmentation into a dimension word vector and a text word vector;
the word vector fusion module is used for calculating the attention score of the text word vector by using an attention mechanism in the trained text emotion classification model and fusing the attention score and the text word vector to obtain a text fusion word vector;
the vector inner product module is used for calculating a vector inner product matrix of the text fusion word vector and the dimension word vector by utilizing a conversion layer in a trained text sentiment classification model, and marking the vector weight of each inner product vector in the vector inner product matrix;
the pooling processing module is used for re-fusing the vector weight and the text fused word vector to obtain a target fused word vector, and pooling the target fused word vector by using a pooling layer in the trained text emotion classification model to obtain a pooled fused word vector;
and the emotion classification judging module is used for calculating the text emotion classification probability of the pooling fusion word vector by using a multi-layer perceptron in the trained text emotion classification model, and determining the emotion classification of the comment text by using an output layer in the trained text emotion classification model according to the text emotion classification probability.
9. An electronic device, characterized in that the electronic device comprises:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,
the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the method of sentiment intelligence classification of comment text as claimed in any one of claims 1 to 7.
10. A computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, implements the method for emotion intelligent classification of comment text according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210961855.5A CN115309864A (en) | 2022-08-11 | 2022-08-11 | Intelligent sentiment classification method and device for comment text, electronic equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210961855.5A CN115309864A (en) | 2022-08-11 | 2022-08-11 | Intelligent sentiment classification method and device for comment text, electronic equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115309864A true CN115309864A (en) | 2022-11-08 |
Family
ID=83860320
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210961855.5A Pending CN115309864A (en) | 2022-08-11 | 2022-08-11 | Intelligent sentiment classification method and device for comment text, electronic equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115309864A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116756326A (en) * | 2023-08-18 | 2023-09-15 | 杭州光云科技股份有限公司 | Emotion and non-emotion text feature analysis and judgment method and device and electronic equipment |
| CN117131347A (en) * | 2023-10-25 | 2023-11-28 | 上海为旌科技有限公司 | Method and device for generating driver dynamic image, electronic equipment and storage medium |
-
2022
- 2022-08-11 CN CN202210961855.5A patent/CN115309864A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116756326A (en) * | 2023-08-18 | 2023-09-15 | 杭州光云科技股份有限公司 | Emotion and non-emotion text feature analysis and judgment method and device and electronic equipment |
| CN116756326B (en) * | 2023-08-18 | 2023-11-24 | 杭州光云科技股份有限公司 | Emotion and non-emotion text feature analysis and judgment method and device and electronic equipment |
| CN117131347A (en) * | 2023-10-25 | 2023-11-28 | 上海为旌科技有限公司 | Method and device for generating driver dynamic image, electronic equipment and storage medium |
| CN117131347B (en) * | 2023-10-25 | 2024-01-19 | 上海为旌科技有限公司 | Method and device for generating driver dynamic image, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Swathi et al. | An optimal deep learning-based LSTM for stock price prediction using twitter sentiment analysis | |
| CN113822494B (en) | Risk prediction method, device, equipment and storage medium | |
| Ain et al. | Sentiment analysis using deep learning techniques: a review | |
| CN112597312A (en) | Text classification method and device, electronic equipment and readable storage medium | |
| CN109923557A (en) | Use continuous regularization training joint multitask neural network model | |
| CN110096575B (en) | Psychological portrait method facing microblog user | |
| CN115309864A (en) | Intelligent sentiment classification method and device for comment text, electronic equipment and medium | |
| CN114648392B (en) | Product recommendation method and device based on user portrait, electronic equipment and medium | |
| CN115392237B (en) | Emotion analysis model training method, device, equipment and storage medium | |
| CN112000778A (en) | Natural language processing method, device and system based on semantic recognition | |
| CN113887930A (en) | Question-answering robot health degree evaluation method, device, equipment and storage medium | |
| CN113627797A (en) | Image generation method and device for employee enrollment, computer equipment and storage medium | |
| Kashif | Urdu Handwritten Text Recognition Using ResNet18 | |
| CN113254814A (en) | Network course video labeling method and device, electronic equipment and medium | |
| Zhang et al. | Improved human-object interaction detection through skeleton-object relations | |
| CN116340516A (en) | Entity relation cluster extraction method, device, equipment and storage medium | |
| Nouhaila et al. | Arabic sentiment analysis based on 1-D convolutional neural network | |
| CN116089605A (en) | Text emotion analysis method based on transfer learning and improved word bag model | |
| Tannert et al. | FlowchartQA: the first large-scale benchmark for reasoning over flowcharts | |
| CN112463966B (en) | False comment detection model training method, false comment detection model training method and false comment detection model training device | |
| CN115510188A (en) | Text keyword association method, device, equipment and storage medium | |
| CN114943306A (en) | Intention classification method, device, equipment and storage medium | |
| Satirapiwong et al. | Information extraction for different layouts of invoice images | |
| CN114911940A (en) | Text emotion recognition method and device, electronic equipment and storage medium | |
| CN113869068A (en) | Scene service recommendation method, device, equipment and storage medium |
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 |