WO2020199600A1

WO2020199600A1 - Sentiment polarity analysis method and related device

Info

Publication number: WO2020199600A1
Application number: PCT/CN2019/118447
Authority: WO
Inventors: 王健宗; 贾雪丽
Original assignee: 平安科技（深圳）有限公司
Priority date: 2019-04-03
Filing date: 2019-11-14
Publication date: 2020-10-08
Also published as: CN110083829A

Abstract

A sentiment polarity analysis method and a related device, applicable in an electronic device (400), comprising: when a coded text-containing target sentence is detected, setting the target sentence as a first input and inputting to a target neural network model; acquiring a first sample sentence from a database, setting the first sample sentence as a second input and inputting to the target neural network model; acquiring an output result of the target neural network, the output result being produced when the target neural network processes the first input and the second input, when the output result is detected to be a first result, determining the first sample sentence as a target sample sentence; extracting a sentiment identifier of the target sample sentence, and determining, on the basis of the sentiment identifier, the sentiment that the target sentence represents. The present solution favors the accurate and rapid determination of the sentiment expressed by a coded text-containing sentence.

Description

Emotion polarity analysis method and related device

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office, the application number is 201910267765.4, and the application name is "Affective Polarity Analysis Method and Related Devices" on April 3, 2019. The entire content is incorporated into this application by reference. in.

Technical field

This application relates to the field of electronic technology, and in particular to an emotional polarity analysis method and related devices.

Background technique

Emotional polarity analysis is a common application of natural language processing methods, especially in classification methods that aim to extract the emotional content of text. In this way, sentiment polarity analysis can be seen as a way to quantify qualitative data using some sentiment score indicators. Although emotions are largely subjective, there are already many useful practices for quantitative emotion analysis, such as companies analyzing consumer feedback on products, or detecting bad reviews in online reviews.

Among them, the simplest sentiment analysis method is to use the positive and negative attributes of words to determine. Each word in the sentence has a score, optimistic words are scored +1, and pessimistic words are scored -1. Then we add up the scores of all words in the sentence to get a final emotional total score. Obviously, this method has many limitations. The most important point is that it ignores contextual information. For example, in this simple model, because "not" has a score of -1 and "good" has a score of +1, the phrase "not good" will be classified as a neutral phrase. But "not good" is usually negative. Another common method is to treat the text as a "bag of words". We see each text as a 1xN vector, where N represents the number of text vocabulary. Each column in the vector is a word, and its corresponding value is the frequency of the word. For example, the phrase "bag of bag of words" can be coded as [2,2,1]. These data can be applied to machine learning classification algorithms (such as Logis regression or support vector machines) to predict the emotional state of unknown data. It should be noted that this supervised learning method requires the use of data with known emotional conditions as a training set. Although this method improves the previous model, it still ignores the context information and the size of the data set.

Summary of the invention

The embodiments of the present application provide an emotional polarity analysis method and related devices, which facilitate the analysis of the emotional polarity of the target sentence paragraph, thereby helping the user to quickly obtain the emotional polarity represented by the target sentence paragraph.

In the first aspect, an embodiment of the present application provides an emotional polarity analysis method, applied to an electronic device, and the method includes:

When the sentiment polarity analysis operation for the target sentence paragraph is detected, acquiring multiple words of the target sentence paragraph;

Input the multiple vocabularies into a first neural network model to obtain a first set of word vectors corresponding to the multiple vocabularies, where the first neural network model is a set of related models used to generate word vectors and is based on Stating a target vocabulary in a plurality of vocabulary and vocabulary adjacent to the target vocabulary to determine a word vector of the target vocabulary, and each word vector in the first word vector set is used to indicate context information of a corresponding vocabulary;

Each word vector in the first word vector set is input to the second neural network model to obtain an output label associated with the first word vector set, and the output label is a label in a preset label set, so Each label in the preset label set is used to indicate an emotional polarity;

The emotional polarity of the target sentence paragraph is determined according to the output tag.

In a second aspect, an embodiment of the present application provides an emotional polarity analysis device, which is applied to an electronic device. The emotional polarity analysis device includes a detection unit, a processing unit, and a determination unit, wherein,

The detection unit is configured to obtain multiple words of the target sentence paragraph when an emotion polarity analysis operation for the target sentence paragraph is detected;

The processing unit is configured to input the multiple words into a first neural network model to obtain a first word vector set corresponding to the multiple words, wherein the first neural network model is used to generate word vectors Related model sets, and determine the word vector of the target vocabulary according to the target vocabulary in the plurality of vocabularies and vocabulary adjacent to the target vocabulary, and each word vector in the first word vector set is used for Indicates the context information of the corresponding vocabulary;

The processing unit is further configured to input each word vector in the first word vector set into a second neural network model to obtain an output label associated with the first word vector set, and the output label is a preset Tags in the tag set of, each tag in the preset tag set is used to indicate an emotional polarity;

The determining unit is configured to determine the emotional polarity of the target sentence paragraph according to the output tag.

In a third aspect, embodiments of the present application provide an electronic device, including a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be processed by the above The above-mentioned program includes instructions for executing the steps in any method of the first aspect of the embodiments of the present application.

In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the foregoing computer-readable storage medium stores a computer program for electronic data exchange, wherein the foregoing computer program enables a computer to execute In one aspect, some or all of the steps described in any method.

In a fifth aspect, embodiments of the present application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute For example, some or all of the steps described in any method of the first aspect. The computer program product may be a software installation package.

It can be seen that in the embodiment of the present application, the electronic device can first obtain the word vector set corresponding to each sentence through the first neural network model when performing emotional polarity analysis on the target sentence paragraph, and then through the second neural network model. The network model obtains the sentiment polarity corresponding to each sentence, and whether the sentiment polarity analysis is performed to analyze a vocabulary separately, and also combines the context corresponding to the vocabulary, which helps to improve the accuracy of sentiment polarity analysis and helps users Quickly get the result of sentiment polarity analysis of the target sentence paragraph.

Description of the drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background art, the following will describe the drawings that need to be used in the embodiments of the present application or the background art.

FIG. 1A is a schematic flowchart of an emotional polarity analysis method provided by an embodiment of the present application;

FIG. 1B is a schematic diagram of the processing flow of a neural network model provided by an embodiment of the present application;

2 is a schematic flowchart of another emotion polarity analysis method provided by an embodiment of the present application;

3 is a schematic flowchart of another emotion polarity analysis method provided by an embodiment of the present application;

4 is a schematic structural diagram of an electronic device provided by an embodiment of the present application;

Fig. 5 is a block diagram of functional units of an emotional polarity analysis device provided by an embodiment of the present application.

detailed description

In order to enable those skilled in the art to better understand the solution of the application, the technical solutions in the embodiments of the application will be clearly and completely described below in conjunction with the drawings in the embodiments of the application. Obviously, the described embodiments are only It is a part of the embodiments of this application, not all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work should fall within the protection scope of this application.

Detailed descriptions are given below.

The terms "first", "second", "third" and "fourth" in the description and claims of the application and the drawings are used to distinguish different objects, rather than describing a specific order . In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes unlisted steps or units, or optionally also includes Other steps or units inherent to these processes, methods, products or equipment.

Reference to "embodiments" herein means that a specific feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art clearly and implicitly understand that the embodiments described herein can be combined with other embodiments.

The following describes the embodiments of the present application in detail.

Please refer to FIG. 1A. FIG. 1A is a schematic flow chart of an emotional polarity analysis method provided by an embodiment of the present application, which is applied to an electronic device. The emotional polarity analysis method includes:

S101: The electronic device acquires multiple vocabulary of the target sentence paragraph when detecting the emotion polarity analysis operation for the target sentence paragraph.

Among them, the sentiment polarity analysis method in this application is suitable for a variety of application scenarios, for example, the sentiment polarity analysis of Taobao product reviews, the sentiment polarity analysis of Weibo comments, and the sentiment polarity analysis of opinion letters for enterprises. When the user's sentiment polarity analysis operation for the target sentence paragraph is detected, multiple words of the target sentence paragraph are obtained. The target sentence paragraph may contain one sentence or multiple sentences. When the target paragraph includes multiple sentences, First determine the emotional polarity corresponding to each sentence in turn, and then determine the emotional polarity of the target sentence paragraph.

S102. The electronic device inputs the multiple words into a first neural network model to obtain a first word vector set corresponding to the multiple words, wherein the first neural network model is a correlation used to generate word vectors Model set, and determine the word vector of the target vocabulary according to the target vocabulary in the plurality of vocabularies and vocabulary adjacent to the target vocabulary, each word vector in the first word vector set is used to indicate The context information of the corresponding vocabulary.

Wherein, after splitting each sentence in the target sentence paragraph, multiple words are obtained, and the multiple words are input into the first neural network model to obtain the corresponding first word vector set, and each word vector includes the corresponding The contextual information of the vocabulary. For example, when a sentence is “Student Xiao Ming’s test score is not very outstanding”, the emotional polarity corresponding to the word “outstanding” is positive, but the sentence is actually negative. Therefore, The word vector corresponding to "prominent" also includes the context information of the word, so that after each vocabulary is converted into a word vector, it is helpful to be more prepared to judge the emotional polarity corresponding to each vocabulary.

S103. The electronic device inputs each word vector in the first word vector set into a second neural network model to obtain an output label associated with the first word vector set, where the output label is a preset label The tags in the set, and each tag in the preset tag set is used to indicate an emotional polarity.

Among them, after each word vector in the first word vector set is input into the second neural network model, an output label associated with the first word vector set can be obtained, and the output label is used to indicate the emotional polarity represented by the target sentence paragraph. This will help Taobao sellers quickly count the number of positive reviews and negative reviews in a large number of user reviews, eliminating the need to read and understand each review.

S104: The electronic device determines the emotional polarity of the target sentence paragraph according to the output tag.

Among them, multiple word vectors corresponding to the target sentence paragraph are obtained through the first neural network model, and then output labels corresponding to the target sentence paragraphs are obtained according to the multiple word vectors through the second neural network model, and the sentiment of the target sentence paragraph can be determined according to the output labels polarity. Since the word vector can be used to indicate the context information of the corresponding vocabulary, the emotional polarity of the target sentence paragraph can be determined more accurately through the first neural network model and the second neural network model. In addition, the first neural network model and the second neural network model are used. Before the neural network model, it is necessary to train the first neural network model and the second neural network model by using a large number of sentences representing positive emotions and a large number of sentences representing negative emotions. The parameters of is adjusted so that the first output label can be output by the input sentence representing positive emotion, and the second output label can be output by the input sentence representing negative emotion.

In a possible example, the inputting the multiple words into the first neural network model to obtain the first word vector set corresponding to the multiple words includes: obtaining multiple sentences of the target paragraph; Each sentence in the multiple sentences is split, and the part of speech of the multiple words obtained after the splitting is determined; a plurality of words whose part of speech is a preset part of speech are selected and input into the first neural network model.

Among them, when the target sentence paragraph has multiple paragraph sentences, first determine the multiple sentences that make up the target sentence paragraph, and then split each sentence. The split sentence consists of multiple words. These words may include nouns, Part of speech vocabulary such as verbs, adjectives, adverbs, prepositions, pronouns, etc., select the words whose part of speech is the preset part of speech among the multiple vocabularies obtained after splitting. For example, the target paragraph sentence is the evaluation of a Taobao buyer. When obtaining multiple words in the target paragraph sentence, the words can be two-character words, three-character words, etc. By selecting preset words of speech, you can Omitting conjunctions or prepositions, such as "的", "地" and punctuation marks, will help to more accurately determine whether the buyer's evaluation is a commendatory evaluation or a derogatory evaluation.

It can be seen that in this example, the part of speech of the vocabulary obtained after splitting is judged, so that multiple words whose part of speech is the preset part of speech are selected and input into the first neural network model, and the prepositions and conjunctions that cannot be judged for emotional polarity are removed Such vocabulary, or emoticons or special characters doped in sentences, help simplify vocabulary composition and improve processing efficiency.

In a possible example, the first neural network model is the Word2vecc neural network model; the inputting the multiple words into the first neural network model to obtain the first word vector set corresponding to the multiple words includes : Convert the multiple vocabularies into one-hot vectors by encoding the multiple vocabularies, the one-hot vectors corresponding to the multiple vocabularies form a second word vector set; combine the second word vector set Each one-hot vector of is sequentially input to the Word2vecc neural network model to obtain the first set of word vectors.

Among them, Word2vecc is a group of related models used to generate word vectors. These models are shallow and two-layer neural networks that are used to train to reconstruct the text of linguistic words. By determining the polarity of a word, it is also necessary to determine the words adjacent to the word. Under the assumption of the word bag model in Word2vecc, the order of the words is not important. After the training is completed, the Word2vecc model can be used to map each word to a vector, which can be used to represent the relationship between word-to-word, and this vector is the hidden layer of the neural network. In addition, word vectors have good semantic characteristics and are a common way to express word features. The value of each dimension of the word vector represents a feature with a certain semantic and grammatical interpretation. Therefore, each dimension of the word vector can be called a word feature. The word vector has many forms, such as one-hot vector. The one-hot vector is used as the input of Word2vecc, and the low-dimensional word vector is trained through Word2vecc.

Among them, the input of Word2vec is also a word vector. Therefore, it is necessary to first convert multiple words into one-hot vectors by encoding to obtain the second word vector set, and each word in the second word vector set corresponds to one -After the hot vector is input into the Word2vec neural network model in turn, the first word vector set is obtained. Word2vec is a way to obtain the second word vector set. The word vector obtained through this way contains possible contextual information. In sentiment analysis Can more accurately recognize the emotion of the text.

Among them, in the input layer of the Word2vec neural network model, the input is a word vector, for example, a word can be converted into a one-hot vector. When training the Word2vec neural network model, the input is a one-hot vector, a one-hot vector means a vector is used to represent a word, and there are N words in this corpus. The dimension of this vector is 1*N, only The element at the corresponding position is 1, and the elements at other positions are all 0. In addition, the output is the vector of the context word of the input word. The elements in the vector are between 0-1, and the one-hot vector that should have appeared takes the cross-entropy as the loss function, and uses backpropagation to train the first A neural network, and the weight matrix of the first neural network.

Among them, after the first neural network model is trained, input the one-hot vector of a vocabulary, and the context word probability of the corresponding vocabulary will be output. At the same time, the first neural network model contains a hidden layer neural network, which trains the input weights. Matrix. Each row of the weight matrix corresponds to the word vector of the word at the corresponding position. After the first neural network model obtains the first word vector set, it is used as the input of the second neural network model.

It can be seen that in this example, the first neural network model can be the Word2vec neural network model. After the target sentence paragraph is split into multiple words, it is first transformed into a second word vector set, and then the first neural network model is used to obtain the first The word vector set, because the word vector in the first word vector set contains possible contextual information, the emotion expressed in the target sentence paragraph can be more accurately identified in sentiment analysis.

In a possible example, the second neural network model is an SVM neural network model.

Among them, when the first neural network model is the Word2vecc neural network model, the second neural network model can be a support vector machine (Support Vector Machine, SVM) model. The SVM model is used to implement sentiment classification and is obtained through the Word2vec neural network model The word vector is used as input and the output is label, which means that the target sentence paragraph is 0 or 1, respectively representing whether the emotional polarity of the target sentence paragraph is positive or negative. The training set of SVM is the word vector representation of a paragraph. To train the SVM model on the training set, the emotional polarity of a paragraph can be judged through the trained model. Therefore, the second neural network model can be the SVM model.

It can be seen that in this example, before inputting multiple words into the Word2vecc neural network model, the words need to be converted into one-hot vectors, and then the second word vector set is obtained through the Word2vecc neural network model. Each word in the second word vector set The vector includes the corresponding context information. After each word vector in the first word vector set is input into the SVM neural network model, the sentiment polarity can be judged through the output label, which is conducive to more accurately inferring the sentiment polarity of the target sentence paragraph .

In a possible example, the output tag includes a first tag and a second tag; the determining the emotional polarity of the target sentence paragraph according to the output tag includes: detecting that the output tag is the In the case of the first tag, determine that the emotional polarity corresponding to the target sentence paragraph is a positive emotion; or, when it is detected that the output tag is the second tag, determine that the emotional polarity corresponding to the target sentence paragraph is Negative emotions.

It can be seen that in this example, after each word vector in the first word vector set is input to the second neural network model, the output label associated with the first word vector set is obtained. Through the output label, the emotional extreme of the target sentence paragraph can be determined. Sex.

In a possible example, after determining the emotional polarity of the target sentence paragraph according to the output tag, the method further includes: displaying the emotional polarity in a preset display area of the target sentence paragraph.

Among them, after the emotional polarity of the target sentence paragraph is obtained, the obtained emotional polarity can be displayed in the preset display area of the target sentence paragraph. The preset display area may be located on the left, right, upper side of the target sentence paragraph. Positions such as the lower side, or the display area on the top of the target sentence paragraph, can be set by the user.

It can be seen that in this example, by displaying the emotional polarity of the target sentence paragraph in the preset display area, a visual display interface is formed, so that the user can quickly obtain the emotional polarity expressed by the target sentence paragraph, which is beneficial to Taobao sellers, etc. Quickly determine the positive and negative reviews of multiple user reviews.

In a possible example, the displaying the emotional polarity in the preset display area of the target sentence paragraph includes: when it is detected that the emotional polarity is a positive emotion, displaying in a first color; Or, when an emotion with a negative emotion polarity is detected, the second color is used for display.

Among them, by outputting tags, it can be determined whether the emotional polarity of the target sentence paragraph is positive or negative, and then different emotional polarities can be displayed differentiated. For example, when it is detected that the emotional polarity of a certain evaluation is positive, that is, the evaluation is favorable, it will be displayed in green. When the emotional polarity of a certain evaluation is negative, that is, the evaluation is negative, it will be displayed in red. The display is performed so that the user can quickly determine whether the Taobao evaluation is a good or bad evaluation.

It can be seen that in this example, when the emotional polarity of the target sentence paragraph is displayed in the preset display area, if the emotional polarity is detected as a positive emotion, the first color is used for display, and the emotional polarity is detected. When sex is a negative emotion, a second color is used for display, and the first color is different from the second color, and different emotional polarities are displayed differentially, which is helpful for users to quickly make judgments.

Consistent with the embodiment shown in FIG. 1A, please refer to FIG. 2. FIG. 2 is a schematic flowchart of an emotional polarity analysis method provided by an embodiment of the present application, which is applied to an electronic device. As shown in the figure, this sentiment polarity analysis method includes:

S201: The electronic device acquires multiple vocabulary of the target sentence paragraph when detecting the emotion polarity analysis operation for the target sentence paragraph.

S202: The electronic device converts the multiple vocabularies into one-hot vectors by encoding the multiple vocabularies, and the one-hot vectors corresponding to the multiple vocabularies form a second word vector set.

S203. The electronic device inputs each one-hot vector in the second word vector set into the Word2vecc neural network model in turn to obtain the first word vector set, and each one-hot vector in the first word vector set A word vector is used to indicate the context information of the corresponding vocabulary.

S204: The electronic device inputs each word vector in the first word vector set into a second neural network model to obtain an output label associated with the first word vector set.

S205: The electronic device determines the emotional polarity of the target sentence paragraph according to the output tag.

In addition, the first neural network model can be the Word2vec neural network model. After the target sentence paragraph is split into multiple words, it is first transformed into a second word vector set, and then the first word vector set is obtained through the first neural network model. Since the word vectors in the first word vector set contain possible contextual information, the emotion expressed in the target sentence paragraph can be more accurately identified in sentiment analysis.

Consistent with the embodiments shown in FIG. 1A and FIG. 2, please refer to FIG. 3. FIG. 3 is a schematic flowchart of an emotional polarity analysis method provided by an embodiment of the present application, which is applied to an electronic device. As shown in the figure, this sentiment polarity analysis method includes:

S301: The electronic device acquires multiple vocabulary of the target sentence paragraph when detecting the emotion polarity analysis operation for the target sentence paragraph.

S302. The electronic device converts the multiple vocabularies into one-hot vectors by encoding the multiple vocabularies, and the one-hot vectors corresponding to the multiple vocabularies form a second word vector set.

S303. The electronic device inputs each one-hot vector in the second word vector set into the Word2vecc neural network model in turn to obtain the first word vector set, wherein, in the first word vector set Each word vector of is used to indicate the context information of the corresponding word.

S304: The electronic device inputs each word vector in the first word vector set into a second neural network model to obtain an output label associated with the first word vector set.

S305: The electronic device determines the emotional polarity of the target sentence paragraph according to the output tag.

S306: The electronic device displays the emotional polarity in a preset display area of the target sentence paragraph.

In addition, by displaying the emotional polarity of the target sentence paragraph in the preset display area, a visual display interface is formed, so that the user can quickly obtain the emotional polarity expressed by the target sentence paragraph, so that Taobao sellers can quickly determine the amount of emotion. Positive and negative reviews in user reviews.

Consistent with the embodiments shown in FIG. 1A, FIG. 2, and FIG. 3, please refer to FIG. 4. FIG. 4 is a schematic structural diagram of an electronic device 400 provided by an embodiment of the present application. The electronic device 400 has one or Multiple application programs and operating systems. As shown in the figure, the electronic device 400 includes a processor 410, a memory 420, a communication interface 430, and one or more programs 421, wherein the one or more programs 421 are stored in the In the memory 420 and configured to be executed by the processor 410, the one or more programs 421 include instructions for executing the following steps;

In a possible example, in terms of inputting the plurality of words into the first neural network model to obtain the first word vector set corresponding to the plurality of words, the instructions in the program are specifically used to perform the following operations : Obtain multiple sentences of the target paragraph; split each sentence in the multiple sentences, and determine the part of speech of the multiple words obtained after the split; select multiple words whose part of speech is preset Input the first neural network model.

In a possible example, the first neural network model is the Word2vecc neural network model; in terms of inputting the multiple words into the first neural network model to obtain the first word vector set corresponding to the multiple words, The instructions in the program are specifically used to perform the following operations: the multiple words are converted into one-hot vectors by encoding the multiple words, and the one-hot vectors corresponding to the multiple words form a second word vector Set; each one-hot vector in the second word vector set is sequentially input into the Word2vecc neural network model to obtain the first word vector set.

In a possible example, where the output tag includes a first tag and a second tag; in terms of determining the emotional polarity of the target sentence paragraph according to the output tag, the instructions in the program are specifically used to execute The following operations: when it is detected that the output tag is the first tag, determine that the emotion polarity corresponding to the target sentence paragraph is a positive emotion; or, when it is detected that the output tag is the second tag , It is determined that the emotion polarity corresponding to the target sentence paragraph is a negative emotion.

In a possible example, after the emotional polarity of the target sentence paragraph is determined according to the output tag, the instructions in the program are specifically used to perform the following operations: in the preset display area of the target sentence paragraph Show the emotional polarity.

In a possible example, in terms of displaying the emotional polarity in the preset display area of the target sentence paragraph, the instructions in the program are specifically used to perform the following operations: after detecting that the emotional polarity is positive In the case of emotion, the first color is used for display; or, when the emotion with the negative emotion polarity is detected, the second color is used for display, and the first color is different from the second color.

The foregoing embodiment mainly introduces the solution of the embodiment of the present application from the perspective of the execution process on the method side. It can be understood that, in order to implement the above-mentioned functions, the electronic device includes hardware structures and/or software modules corresponding to each function. Those skilled in the art should easily realize that in combination with the units and algorithm steps of the examples described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed by hardware or computer software-driven hardware depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.

The embodiment of the present application may divide the electronic device into functional units according to the method example. For example, each functional unit may be divided corresponding to each function, or two or more functions may be integrated into one processing unit. The integrated unit can be implemented in the form of hardware or software functional unit. It should be noted that the division of units in the embodiments of the present application is illustrative, and is only a logical function division, and there may be other division methods in actual implementation.

The following are device embodiments of the present invention, and the device embodiments of the present invention are used to execute the methods implemented in the method embodiments of the present invention. The emotion polarity analysis device 500 shown in FIG. 5 is applied to the electronic device. The emotion polarity analysis device includes a detection unit 501, a processing unit 502, and a determination unit 503, wherein,

The detection unit 501 is configured to acquire multiple vocabulary of the target sentence paragraph when the emotion polarity analysis operation for the target sentence paragraph is detected;

The processing unit 502 is configured to input the plurality of words into a first neural network model to obtain a first set of word vectors corresponding to the plurality of words, wherein the first neural network model is used to generate word vectors And determine the word vector of the target vocabulary according to the target vocabulary in the multiple vocabularies and vocabulary adjacent to the target vocabulary, and each word vector in the first word vector set is used To indicate the contextual information of the corresponding vocabulary;

The processing unit 502 is further configured to input each word vector in the first word vector set into a second neural network model to obtain an output label associated with the first word vector set, and the output label is a pre- Tags in the set of tags, each tag in the preset tag set is used to indicate an emotional polarity;

The determining unit 503 is configured to determine the emotional polarity of the target sentence paragraph according to the output tag.

In a possible example, in the aspect of inputting the multiple words into the first neural network model to obtain the first word vector set corresponding to the multiple words, the processing unit 503 is further configured to: obtain the Multiple sentences of the target paragraph; and used to split each of the multiple sentences, and determine the part of speech of the multiple words obtained after the split; and used to select multiple parts of speech as the preset part of speech Vocabulary and input the first neural network model.

In a possible example, the first neural network model is the Word2vecc neural network model; the first neural network model is the Word2vecc neural network model; and the multiple words are input into the first neural network model to obtain Regarding the first word vector set corresponding to the multiple words, the processing unit 503 is specifically configured to: convert the multiple words into one-hot vectors by encoding the multiple words, and the multiple words The corresponding one-hot vectors form a second word vector set; and are used to input each one-hot vector in the second word vector set into the Word2vecc neural network model in turn to obtain the first word vector set.

In a possible example, where the output tags include a first tag and a second tag; in terms of determining the emotional polarity of the target sentence paragraph according to the output tags, the processing unit 503 is specifically configured to: When it is detected that the output tag is the first tag, it is determined that the emotional polarity corresponding to the target sentence paragraph is a positive emotion; or, when it is detected that the output tag is the second tag, it is determined that the The emotion polarity corresponding to the target sentence paragraph is negative emotion.

In a possible example, after the emotion polarity of the target sentence paragraph is determined according to the output tag, the processing unit 503 is specifically configured to: display the emotion in a preset display area of the target sentence paragraph polarity.

In a possible example, in terms of displaying the emotion polarity in the preset display area of the target sentence paragraph, the processing unit 503 is specifically configured to: when detecting that the emotion polarity is a positive emotion, use The first color is used for display; or, when the negative emotion of the emotional polarity is detected, the second color is used for display, and the first color is different from the second color.

An embodiment of the present application also provides a computer storage medium, wherein the computer storage medium stores a computer program for electronic data exchange, and the computer program enables a computer to execute part or all of the steps of any method as recorded in the above method embodiment , The aforementioned computer includes electronic equipment.

The embodiments of the present application also provide a computer program product. The above-mentioned computer program product includes a non-transitory computer-readable storage medium storing a computer program. The above-mentioned computer program is operable to cause a computer to execute any of the methods described in the above-mentioned method embodiments. Part or all of the steps of the method. The computer program product may be a software installation package, and the above-mentioned computer includes electronic equipment.

It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should know that this application is not limited by the described sequence of actions. Because according to this application, some steps can be performed in other order or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

In the above-mentioned embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in an embodiment, reference may be made to related descriptions of other embodiments.

In the several embodiments provided in this application, it should be understood that the disclosed device may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the above-mentioned units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components can be combined or integrated. To another system, or some features can be ignored, or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical or other forms.

The units described above as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.

In addition, the functional units in each embodiment of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or software functional unit.

If the above integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable memory. Based on this understanding, the technical solution of the present application essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, A number of instructions are included to enable a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the foregoing methods of the various embodiments of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other various media that can store program codes.

Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above-mentioned embodiments can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable memory, and the memory can include: flash disk , Read-only memory (English: Read-Only Memory, abbreviation: ROM), random access device (English: Random Access Memory, abbreviation: RAM), magnetic disk or optical disc, etc.

The embodiments of the application are described in detail above, and specific examples are used in this article to illustrate the principles and implementation of the application. The descriptions of the above examples are only used to help understand the methods and core ideas of the application; A person of ordinary skill in the art, based on the idea of the present application, will have changes in the specific implementation and the scope of application. In summary, the content of this specification should not be construed as a limitation of the present application.

Claims

An emotional polarity analysis method, characterized in that it is applied to an electronic device, and the method includes:

When the sentiment polarity analysis operation for the target sentence paragraph is detected, acquiring multiple words of the target sentence paragraph;

Input the multiple vocabularies into a first neural network model to obtain a first set of word vectors corresponding to the multiple vocabularies, where the first neural network model is a set of related models used to generate word vectors and is based on Stating a target vocabulary in a plurality of vocabulary and vocabulary adjacent to the target vocabulary to determine a word vector of the target vocabulary, and each word vector in the first word vector set is used to indicate context information of a corresponding vocabulary;

Each word vector in the first word vector set is input to the second neural network model to obtain an output label associated with the first word vector set, and the output label is a label in a preset label set, so Each label in the preset label set is used to indicate an emotional polarity;

The emotional polarity of the target sentence paragraph is determined according to the output tag.
The method according to claim 1, wherein said inputting said plurality of words into a first neural network model to obtain a first set of word vectors corresponding to said plurality of words comprises:

Acquiring multiple sentences of the target paragraph;

Split each sentence in the plurality of sentences, and determine the part of speech of the plurality of words obtained after the split;

A plurality of words whose part of speech is a preset part of speech are selected and input into the first neural network model.
The method of claim 1 or 2, wherein the first neural network model is a Word2vecc neural network model; and the multiple words are input into the first neural network model to obtain the corresponding words The first word vector set includes:

Converting the multiple vocabularies into one-hot vectors by encoding the multiple vocabularies, and one-hot vectors corresponding to the multiple vocabularies form a second word vector set;

Each one-hot vector in the second word vector set is sequentially input into the Word2vecc neural network model to obtain the first word vector set.
The method according to any one of claims 1 to 3, wherein the second neural network model is an SVM neural network model.
The method of claim 1, wherein the output tag includes a first tag and a second tag; and the determining the emotional polarity of the target sentence paragraph according to the output tag includes:

When it is detected that the output tag is the first tag, it is determined that the emotion polarity corresponding to the target sentence paragraph is a positive emotion; or,

When it is detected that the output tag is the second tag, it is determined that the emotion polarity corresponding to the target sentence paragraph is a negative emotion.
5. The method according to any one of claims 1 to 5, wherein after the determining the emotional polarity of the target sentence paragraph according to the output label, the method further comprises:

The emotional polarity is displayed in the preset display area of the target sentence paragraph.
7. The method of claim 6, wherein the displaying the emotional polarity in the preset display area of the target sentence paragraph comprises:

When detecting that the emotion polarity is a positive emotion, use the first color for display; or,

When the negative emotion of the emotion polarity is detected, a second color is used for display, and the first color is different from the second color.
An emotion polarity analysis device, characterized in that it is applied to electronic equipment, the emotion polarity analysis device includes a detection unit, a processing unit, and a determination unit, wherein:

The detection unit is configured to obtain multiple words of the target sentence paragraph when an emotion polarity analysis operation for the target sentence paragraph is detected;

The processing unit is configured to input the multiple words into a first neural network model to obtain a first word vector set corresponding to the multiple words, wherein the first neural network model is used to generate word vectors Related model sets, and determine the word vector of the target vocabulary according to the target vocabulary in the plurality of vocabularies and vocabulary adjacent to the target vocabulary, and each word vector in the first word vector set is used for Indicates the context information of the corresponding vocabulary;

The processing unit is further configured to input each word vector in the first word vector set into a second neural network model to obtain an output label associated with the first word vector set, and the output label is a preset Tags in the tag set of, each tag in the preset tag set is used to indicate an emotional polarity;

The determining unit is configured to determine the emotional polarity of the target sentence paragraph according to the output tag.
The emotion polarity analysis device according to claim 8, wherein, in the aspect of inputting the plurality of words into the first neural network model to obtain the first word vector set corresponding to the plurality of words, the The processing unit is specifically used for: obtaining multiple sentences of the target paragraph; and for splitting each sentence of the multiple sentences, and determining the parts of speech of the multiple words obtained after splitting; and A plurality of words whose part of speech is a preset part of speech are selected and input into the first neural network model.
The emotion polarity analysis device according to claim 8 or 9, wherein the first neural network model is a Word2vecc neural network model; in the inputting the plurality of words into the first neural network model, the result Regarding the first set of word vectors corresponding to multiple words, the processing unit is specifically configured to: convert the multiple words into one-hot vectors by encoding the multiple words, and the multiple words corresponding to the The one-hot vector forms a second word vector set; and is used to input each one-hot vector in the second word vector set into the Word2vecc neural network model in turn to obtain the first word vector set.
The emotional polarity analysis device according to any one of claims 8-10, wherein the second neural network model is an SVM neural network model.
8. The emotion polarity analysis device according to claim 8, wherein the output tag includes a first tag and a second tag. In terms of determining the emotion polarity of the target sentence paragraph according to the output tag, The determining unit is specifically configured to: when detecting that the output tag is the first tag, determine that the emotion polarity corresponding to the target sentence paragraph is a positive emotion; or, when the output tag is detected When it is the second tag, it is determined that the emotion polarity corresponding to the target sentence paragraph is a negative emotion.
The emotional polarity analysis device according to any one of claims 8-12, wherein after the emotional polarity of the target sentence paragraph is determined according to the output tag, the processing unit is further configured to The preset display area of the target sentence paragraph displays the emotional polarity.
The emotion polarity analysis device according to claim 13, wherein in the aspect of displaying the emotion polarity in the preset display area of the target sentence paragraph, the processing unit is specifically configured to: When the emotional polarity is a positive emotion, the first color is used for display; or when the emotional polarity is negative, the second color is used for display, and the first color is different from the The second color.
An electronic device, characterized by comprising a processor and a memory, the processor and the memory are connected to each other, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to To call the program instructions, perform the following steps:

When the sentiment polarity analysis operation for the target sentence paragraph is detected, acquiring multiple words of the target sentence paragraph;

Inputting the plurality of words into the first neural network model to obtain a first word vector set corresponding to the plurality of words, and each word vector in the first word vector set is used to indicate context information of the corresponding word;

Input each word vector in the first word vector set into a second neural network model to obtain an output label associated with the first word vector set;

The emotional polarity of the target sentence paragraph is determined according to the output tag.
The electronic device according to claim 15, wherein the processor calls the program instructions to execute the input of the plurality of words into the first neural network model to obtain the first word corresponding to the plurality of words When vector collection, perform the following steps:

Acquiring multiple sentences of the target paragraph;

Split each sentence in the plurality of sentences, and determine the part of speech of the plurality of words obtained after the split;

A plurality of words whose part of speech is a preset part of speech are selected and input into the first neural network model.
The electronic device according to claim 15 or 16, wherein the first neural network model is a Word2vecc neural network model; the processor calls the program instructions to execute the input of the multiple words into the first When the neural network model obtains the first word vector set corresponding to the multiple words, the following steps are specifically performed:

Converting the multiple vocabularies into one-hot vectors by encoding the multiple vocabularies, and one-hot vectors corresponding to the multiple vocabularies form a second word vector set;

Each one-hot vector in the second word vector set is sequentially input into the Word2vecc neural network model to obtain the first word vector set.
The electronic device according to any one of claims 15-17, wherein the second neural network model is an SVM neural network model.
The electronic device according to claim 15, wherein the output tag includes a first tag and a second tag; the processor calls the program instructions to execute the determination of the target sentence paragraph based on the output tag In the case of emotional polarity, perform the following steps:

When it is detected that the output tag is the first tag, it is determined that the emotion polarity corresponding to the target sentence paragraph is a positive emotion; or,

When it is detected that the output tag is the second tag, it is determined that the emotion polarity corresponding to the target sentence paragraph is a negative emotion.
A computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program includes program instructions that, when executed by a processor, cause the processor to execute The method of any one of 1-7 is required.