WO2016197767A2

WO2016197767A2 - Method and device for inputting expression, terminal, and computer readable storage medium

Info

Publication number: WO2016197767A2
Application number: PCT/CN2016/081324
Authority: WO
Inventors: 陈军
Original assignee: 中兴通讯股份有限公司
Priority date: 2016-02-16
Filing date: 2016-05-06
Publication date: 2016-12-15
Also published as: WO2016197767A3; CN107423277A

Abstract

Disclosed are a method and device for inputting an expression, a terminal, and a computer readable storage medium. The method comprises: obtaining a session record of an instant communication application, chat content of a first user and another user being recorded in the session record; analysing the session record, and obtaining emotional tendency information of the first user; placing expressions within an expression library which match the emotional tendency information of the first user into an input candidate pool; obtaining in real time real-time chat content inputted by the first user via the instant communication application, and according to the real-time chat content, matching and obtaining from within the input candidate pool a corresponding candidate expression for the first user to select.

Description

Expression input method, device, terminal and computer readable storage medium

Technical field

The present application relates to, but is not limited to, the field of communications, and in particular, to an expression input method, apparatus, terminal, and computer readable storage medium.

Background technique

When a user communicates through a mobile terminal or an Internet device, symbols or pictures with facial expressions, such as character emoticons, Emoji emoticons, emoticons, or animations, are often added to add interactive fun. At present, the way to input or send an emoticon picture or an emoticon is that the user manually selects and searches in the expression library of different topics, and the user selects the corresponding matching string or index word to match the emoticon. Enter and send.

However, find a suitable expression one by one in a large number of expressions in the expression library, or directly input the correct matching keyword, especially when you are in a multi-crowd chat, when you find the desired expression, you will receive more For information, sending an emoticon at this time may be out of place. In the related art, the expression search process is time-consuming and laborious, which causes certain troubles for the user's use, and the user experience is not high.

Summary of the invention

The invention provides an expression input method, device, terminal and computer readable storage medium, which solves the problem that the expression input method in the related art is time consuming and laborious, and the user experience is poor.

An expression input method includes:

Obtain a session record of the instant messaging application, and record the chat content of the first user with other users in the session record.

The session record is analyzed to obtain the emotional tendency information of the first user.

An expression matching the first user's sentiment orientation information in the expression library is placed in the input candidate pool.

The real-time chat content input by the first user through the instant messaging application is obtained in real time, and the corresponding candidate expression is obtained from the input candidate pool according to the real-time chat content for the first user to select.

Optionally, analyzing the session record to obtain the emotional tendency information of the first user includes:

Generate corresponding text data based on the session record.

Semantic analysis of the text data to obtain scene features and/or emotional features of the current session.

The sentiment orientation information of the first user is determined according to the scene feature and/or the emotional feature of the current session.

Optionally, performing semantic analysis on the text data to obtain scene features and/or emotional features of the current session includes:

Semantic analysis of the conversation subject content of the text data to obtain scene features of the current session, wherein different scene features have different scene key values; and/or,

Semantic analysis is performed on the chat content input by the first user in the text data to obtain an emotional feature of the first user, wherein different emotional features have different emotional feature values.

Optionally, generating corresponding text data according to the session record includes:

When an emoticon exists in the session record, the emoticon in the session record is converted into a corresponding tag text;

Generates text data based on the converted session record.

When the voice content exists in the session record, the voice content is identified to obtain a corresponding text;

The voice content in the session record is converted into corresponding text, and the corresponding text data of the session record is generated.

Optionally, the method further includes: prior to the step of placing the expression matching the first user's sentiment orientation information in the input candidate pool, the expressions in the expression database are given respective scene key values and/or emotions. Eigenvalues.

Optionally, the real-time chat content input by the first user through the instant messaging application is obtained in real time, and the corresponding candidate expression is obtained by matching the real-time chat content from the input candidate pool for the first user to select:

Acquiring keywords in real-time chat content input by the first user through the instant messaging application in real time;

Calculating the similarity between the keyword and the tag text or index word of the expression in the input candidate pool;

Matching the expression with the highest similarity to the keyword in the input candidate pool for the first user to select; or,

Determining a contextual scene and/or an emotional tendency based on keywords in the live chat content;

Calculating the similarity between the contextual scene and/or the sentimental tendency and the label text or index term of the expression in the input candidate pool;

Matching the expression with the highest similarity to the context scene in the input candidate pool for the first user to select.

An expression input device comprising:

The obtaining module is configured to obtain a session record of the instant messaging application, and the session record records the chat content of the first user and other users;

The analysis module is configured to analyze the session record to obtain the emotional tendency information of the first user;

a first processing module, configured to place an expression in the expression library that matches the sentiment orientation information of the first user into the input candidate pool;

The second processing module is configured to acquire real-time chat content input by the first user through the instant messaging application in real time, and obtain corresponding candidate expressions for selection by the first user according to the real-time chat content matching from the input candidate pool.

Optionally, the analysis module comprises:

Generating a unit, configured to generate corresponding text data according to the session record;

The analyzing unit is configured to perform semantic analysis on the text data to obtain scene features and/or emotional features of the current session;

The first processing unit is configured to determine the sentiment orientation information of the first user according to the scene feature and/or the emotional feature of the current session.

Optionally, the analysis unit comprises:

The first analysis sub-unit is configured to perform semantic analysis on the conversation subject content of the text data to obtain a scene feature of the current session, wherein different scene features have different scene key values; and/or,

The second analysis subunit is configured to perform semantic analysis on the chat content input by the first user in the text data to obtain an emotional feature of the first user, wherein different emotional features have different emotional feature values.

Optionally, the generating unit includes:

a first conversion subunit, configured to convert an emoticon in the session record into a corresponding label text when an emoticon exists in the session record;

The first generation subunit is configured to generate text data according to the converted session record.

Optionally, the generating unit further includes:

The second conversion subunit is set to enter the voice content when the voice content exists in the session record Line recognition to get the corresponding text;

The second generation subunit is configured to convert the voice content in the session record into corresponding text, and generate corresponding text data of the session record.

Optionally, the expression input device further includes:

The pre-processing module is configured to: after the first processing module puts the expression matching the sentiment orientation information of the first user in the expression library into the input candidate pool, assign the respective scene key values to the expressions in the expression library and/or Or emotional feature values.

Optionally, the second processing module includes:

a first acquiring unit, configured to acquire, in real time, a keyword in the real-time chat content input by the first user through the instant messaging application;

a first calculating unit configured to calculate a similarity between the keyword and the label text or the index word of the expression in the input candidate pool;

a first matching unit, configured to match an expression with the highest similarity of keywords in the input candidate pool for the first user to select; or

a second acquiring unit, configured to acquire, in real time, a keyword in the real-time chat content input by the first user through the instant messaging application;

a second processing unit configured to determine a contextual scene and/or an emotional tendency based on keywords in the live chat content;

a second calculating unit configured to calculate a similarity between the context scene and/or the sentiment orientation and the label text or the index word of the expression in the input candidate pool;

The second matching unit is configured to match the expression with the highest similarity to the context scene in the input candidate pool for the first user to select.

A terminal comprising an expression input device as described above.

A computer readable storage medium storing computer executable instructions that, when executed by a processor, implement the above described expression input method.

The beneficial effects of embodiments of the present invention are:

By analyzing the obtained session record to determine the current user's emotional tendency, all expressions matching the user's emotional tendency are placed in the input candidate pool to reduce the number of candidate expressions and increase the input speed; and according to the user input real-time The chat content matches the corresponding candidate expression, reaching the knot The contextual scene and the user's emotional tendency, so as to quickly find the user's current expression, save time and effort, improve the user experience.

BRIEF abstract

1 is a schematic flow chart of an expression input method according to an embodiment of the present invention;

2 is a schematic flow chart of step S102 in FIG. 1;

Figure 3 is a flow chart 1 of the step S104 of Figure 1;

4 is a second schematic diagram of the process of step S104 in FIG. 1;

FIG. 5 is a schematic structural diagram of an expression input device according to an embodiment of the present invention.

Embodiments of the invention

Embodiments of the present invention will be described below with reference to the accompanying drawings. It should be noted that, in the case of no conflict, the features in the embodiments and the embodiments in the present application may be arbitrarily combined with each other.

Embodiment 1

As shown in FIG. 1 , an embodiment of the present invention provides an expression input method, including steps S101 to S104:

Step S101: Acquire a session record of the instant messaging application.

The session record records the chat content of the first user and other users. The session record may be the chat content between the first user and the single user, or may be the group chat content between the first user and the plurality of users. It is worth noting that in order to ensure the real-time and validity of the session record, it is necessary to select the session record in the previous preset time period at the current time.

Step S102: analyzing the session record to obtain the sentiment orientation information of the first user.

By performing semantic analysis on the session record, the first user's sentiment orientation information can be obtained, that is, the first user's sentiment orientation information can be determined more accurately in combination with the actual chat context scene and the first user's subjective view or feeling. Among them, the use of relevant natural language semantic analysis technology can make the machine understand the human language roughly, so the semantic analysis of the session record is technically feasible.

Step S103: Put an expression in the expression library that matches the emotion tendency information of the first user into the input candidate pool.

The expression library stores the system emoticons and all the expressions downloaded or collected by the first user, and the expressions are The expressions matching the first user's sentiment orientation information in the library are placed in the input candidate pool, which can reduce the number of expressions in the input candidate pool, and even if the user manually finds the expression, the search range can be narrowed and the search efficiency can be improved. It is worth pointing out that the user can ignore the input candidate pool and find the desired expression from the original expression library.

Step S104: Real-time acquisition of the real-time chat content input by the first user through the instant communication application, and matching the real-time chat content from the input candidate pool to obtain the corresponding candidate expression for the first user to select.

The input candidate pool has provided the expression most relevant to the first user's emotional tendency for the user to select, the user can directly browse the selection in the input candidate pool, or can select the system recommended expression by inputting the real-time chat content, where the live chat is At least one index word is included in the content to facilitate expression matching.

By analyzing the obtained session record to determine the current user's emotional tendency, all expressions matching the user's emotional tendency are placed in the input candidate pool to reduce the number of candidate expressions and increase the input speed; and according to the user input real-time The chat content matches the corresponding candidate expressions, and achieves the combination of the context scene and the user's emotional tendency, thereby quickly finding the expression that the user currently needs, saving time and effort, and improving the user experience.

Embodiment 2

The above embodiment 1 briefly introduces the expression input method of the embodiment of the present invention. The following embodiment will introduce the expression input method based on the above embodiment and the application scenario.

As shown in FIG. 2, step S102 includes steps S201 to S203:

Step S201: Generate corresponding text data according to the session record.

Among them, semantic analysis is mainly for text data, but there may be emoticons or voice messages in the session records, so it is necessary to convert the session records into full-text data in order to perform accurate semantic analysis. When an emoticon exists in the session record, the emoticon in the session record is converted into a corresponding tag text; and the text data is generated according to the converted session record. The label text refers to a keyword or phrase used to represent the theme feature of the emoticon when the emoticon is stored in the terminal. When the voice content exists in the session record, the voice content is identified to obtain the corresponding text; the voice content in the session record is converted into the corresponding text, and the corresponding text data of the session record is generated. In the session record, the expression and voice content in the session record are converted into text. All of the content is converted to text data.

Step S202: Perform semantic analysis on the text data to obtain scene features and/or emotion features of the current session.

The main role of the expression in the chat process is to activate the chat atmosphere and express the subjective feelings of the user. Therefore, the appropriate expression is based on a certain scene and the user's own feelings. In order to ensure that the recommended expression is required by the user, the session record is required. The text data is semantically analyzed to obtain scene features and/or emotional features of the current session. Step S202 includes:

Semantic analysis of the text data of the session record can get the context represented by the context. For example, the user and the friend are traveling. In general, the experience of the tour is easy and pleasant, then the scene key value of the big scene is Can be sure. It should be noted that the scene key values corresponding to different scenes are different. For example, the scene can be roughly divided into positive, negative, and neutral. The positive scene key value is higher, and the negative scene key value is lower. The positive, neutral, and negative scene key values are sequentially reduced, while the forward scenes can also be divided into different levels, which are reflected in the scene's key value.

Semantic analysis of the text data of the session record can be used to know the user's subjective feelings. For example, during the chat process, the user usually publishes some words indicating their own opinions or opinions, including emoji, network terms, derogatory words, approval/objection. Use words, or some words that directly express praise, happiness, or words that indicate depression and sadness, in order to obtain the emotional characteristics of the user, wherein different emotional features have different emotional feature values, such as emotions expressing positive emotions. The emotional eigenvalue of the feature is higher than the emotional eigenvalue of the emotional feature representing the negative emotion, and the positive emotion can also be divided into different levels, which are reflected in the level of the emotional feature value.

Step S203: Determine the emotional tendency information of the first user according to the scene feature and/or the emotional feature of the current session.

Wherein, for analyzing the session record only to obtain the scene feature, the user's sentiment tendency information can be directly determined according to the scene feature. For the case where the analysis session record only gets emotional features, it can be straight The user's emotional tendency information is determined based on the emotional characteristics. For the analysis session record, the scene feature of the current session is obtained, and the user's emotional characteristics are obtained, and the user's emotional tendency information can be determined together. After semantic analysis of the session record to obtain the context scene, the scene feature of the current session can be determined, the corresponding scene key value can be obtained, and the user's emotional text in the context can be analyzed to adjust the current scene key value to obtain an accurate emotion. The tendency information, for example, the emotional feature representing the positive emotion may increase the scene key value, and the emotional feature indicating the negative emotion may reduce the scene key value, thereby obtaining accurate emotional tendency information. For example, between the user and the friend is discussing the holiday to the seaside tourism, then the semantic analysis, analysis and analysis of the seaside tourism scene, according to the scene library convention usually gives a pleasant emotional characteristics, and then according to the subjective attitude expressed in the context Enhance or weaken the emotional value. For example, if the user agrees to the view of the seaside tourism in the context, the emotional value is increased. If the user expresses the objection, the emotional value is reduced, and the user's emotional tendency of the context expression is finally determined.

It is worth noting that the emotional eigenvalues of different scenes and different modal and emotional words are stored in the terminal by modeling beforehand, which is convenient for the terminal to quickly read the scene analysis and sentiment analysis. At the same time, the contextual text data is not In violation of user privacy settings, you can also upload to the cloud for big data analysis.

After obtaining the accurate sentiment orientation information, the expression matching the first user's sentiment orientation information in the expression database is placed in the input candidate pool. Before this, the expression input method further includes: assigning respective expressions of the scene key value and/or the emotion feature value to the expressions in the expression library. Each expression of the expression library is classified in advance, for example, into hi, anger, worry, thought, sadness, fear, and shock. In each category, different expressions can be given the same or different scene key values and / or emotional feature values. After obtaining the accurate sentiment orientation information, the expression matching the user's emotional tendency information can be placed in the input candidate pool. If the user wants to actively find an expression, the user can select the contextual emotional feature in the input candidate pool. Expressions, without having to search through all the expression libraries, greatly shorten the search time.

In addition, in addition to actively finding the desired expression, the user can also select the system-recommended expression by inputting the real-time chat content, where the real-time chat content includes at least one index word to facilitate the expression matching. Step S104 shown in FIG. 3 includes steps S301 to S303:

Step S301: Acquire keywords in real-time chat content input by the first user through the instant messaging application in real time.

The input candidate pool provides the user with the expression most relevant to the current conversation scene feature and the user emotion feature, and the user can match the corresponding expression by inputting the label text or index word of the expression to achieve fast search. The user can also match the corresponding expression by inputting keywords that are related or similar to the expression tag text or index words.

Step S302: Calculate the similarity between the keyword and the tag text or the index word of the expression in the input candidate pool.

Since the user may not be able to correctly remember the label text or index word of the expression, a similar but inaccurate keyword may be input, so after obtaining the keyword input by the user, the keyword needs to be input with the expression in the candidate pool. The calculation of the similarity of the label text or index word. The similarity can be calculated by the distance between the word vectors in the semantic analysis method. It is obvious that the input keywords are exactly the same as the expression tags or index words, and the distance between the word vectors is zero; the characters are not identical or the semantics are similar. Then the distance between the word vectors is also small, and the matching degree is also high.

Step S303: Match the expression with the highest similarity with the keyword in the input candidate pool for the first user to select.

In order to prevent the user from inputting similar but unmatched keywords, the simple keyword matching method cannot be used to input the expression. It is necessary to calculate the similarity between the keyword input by the user and the label text or index word of all the expressions in the input candidate pool. And matching the expression with the highest similarity with the keyword in the input candidate pool for the first user to select, so as to optimize the appropriate expression determination input. The matching expressions may be one or more.

Alternatively, as shown in FIG. 4, step S104 further includes steps S401 to S404:

Step S401: Acquire keywords in real-time chat content input by the first user through the instant messaging application in real time.

The input candidate pool provides the user with the expression most relevant to the current conversation scene feature and the user emotion feature, and the user can match the corresponding expression by inputting the label text or index word of the expression to achieve fast search. The user can also match the corresponding expression by inputting a scene keyword related to or similar to the expression label text or the index word.

Step S402: Determine a context scene and/or an emotional tendency according to keywords in the real-time chat content.

Since the same keyword may have different context scenarios, it can be separated by context semantics. The context is determined first and the user's emotional tendency is reflected.

Step S403: Calculate the similarity between the context scene and/or the sentiment orientation and the label text or index word of the expression in the input candidate pool.

Since the user may not be able to correctly remember the label text or index word of the expression, other keywords that are not related to the keyboard may be input. Therefore, after obtaining the keyword input by the user, the context scene reflected by the keyword needs to be// Or the user's emotional tendency is calculated similarly to the tag text or index word of the expression in the input candidate pool. The similarity can be calculated by using the distance between the word vectors in the semantic analysis method. It is obvious that the words representing the scene features and/or the user's emotional features are identical to the expression label text or index words, and the distance of the word vector is zero; Characters are not identical or have similar semantics. Usually the distance between word vectors is smaller and the matching degree is higher.

Step S404: Match the expressions in the input candidate pool with the highest similarity to the context scene and/or the sentiment tendency for the first user to select.

For example, there is an expression in the input candidate pool, and the index word is “Happy Birthday”. According to the strict matching rule, the user must type these four words before the expression appears to be determined by the user. Real-time calculation of the similarity between the word typed by the user and the index word, when the user types "birthday" or "fast birthday", the emoticon can be matched; or, according to the context scene that has been acquired, it has been analyzed Talking about the topic of "birthday", then combining the scene and contextual sentiment, when the user types "happy", the expression expressing happy birthday can be recommended to the user.

In summary, by analyzing the obtained session record to determine the current user's emotional tendency, all expressions matching the user's emotional tendency are placed in the input candidate pool to reduce the number of candidate expressions and increase the input speed; The input real-time chat content matches the corresponding candidate expressions, and achieves the combination of the context scene and the user's emotional tendency, thereby quickly finding the expression that the user currently needs, saving time and effort, and improving the user experience.

In addition, the expression input method provided by the embodiment of the present invention can be applied to the scene of intelligently changing the chat background image or reading the background image in addition to the application and the real-time chat process, so as to improve the user experience.

Embodiment 3

The embodiment 1 and the second embodiment respectively describe the expression input method of the embodiment of the present invention in combination with different application scenarios. The following embodiment will perform the corresponding device with reference to the accompanying drawings. Introduction.

As shown in FIG. 5, another aspect of an embodiment of the present invention further provides an expression input device, including:

The obtaining module 51 is configured to obtain a session record of the instant messaging application, where the chat content of the first user and other users is recorded in the session record;

The analyzing module 52 is configured to analyze the session record to obtain the sentiment orientation information of the first user;

The first processing module 53 is configured to: put an expression in the expression library that matches the sentiment orientation information of the first user into the input candidate pool;

The second processing module 54 is configured to acquire real-time chat content input by the first user through the instant messaging application in real time, and obtain a corresponding candidate expression from the input candidate pool according to the real-time chat content for the first user to select.

Among them, the analysis module includes:

Among them, the analysis unit includes:

The generating unit includes:

The generating unit further includes:

The second conversion subunit is configured to: when the voice content exists in the session record, identify the voice content to obtain a corresponding text;

The expression input device further includes:

The second processing module includes:

It should be noted that the device is a device corresponding to the above-mentioned expression input method, and all the implementation manners in the foregoing method embodiments are applicable to the embodiment of the device, and the same technical effects can be achieved.

According to still another aspect of an embodiment of the present invention, there is also provided a terminal comprising the expression input device as described above. The terminal is a smart device with the function of installing an instant messaging application, such as a smart phone, a tablet or a laptop.

A computer readable storage medium storing computer executable instructions that, when executed by a processor, implement the expression input method.

One of ordinary skill in the art will appreciate that all or a portion of the steps of the above-described embodiments can be implemented using a computer program flow, which can be stored in a computer readable storage medium, such as on a corresponding hardware platform (eg, The system, device, device, device, etc. are executed, and when executed, include one or a combination of the steps of the method embodiments.

Alternatively, all or part of the steps of the above embodiments may also be implemented by using an integrated circuit. These steps may be separately fabricated into individual integrated circuit modules, or multiple modules or steps may be fabricated into a single integrated circuit module. achieve.

The devices/function modules/functional units in the above embodiments may be implemented by a general-purpose computing device, which may be centralized on a single computing device or distributed over a network of multiple computing devices.

When the device/function module/functional unit in the above embodiment is implemented in the form of a software function module and sold or used as a stand-alone product, it can be stored in a computer readable storage medium. The above mentioned computer readable storage medium may be a read only memory, a magnetic disk or an optical disk or the like.

Industrial applicability

The solution of the embodiment of the present invention determines the sentiment tendency of the current user by analyzing the obtained session record, and puts all the expressions matching the user's sentiment tendency into the input candidate pool, thereby reducing the number of candidate expressions and increasing the input speed; According to the real-time chat content input by the user, the corresponding candidate expressions are matched, and the context scene and the user's emotional tendency are combined to quickly find the expression that the user currently needs, saving time and effort, and improving the user experience.

Claims

An expression input method includes:

Obtaining a session record of the instant messaging application, where the chat content of the first user and other users is recorded in the session record;

Performing analysis on the session record to obtain sentiment orientation information of the first user;

Forming an expression in the expression library that matches the sentiment orientation information of the first user into the input candidate pool;

Real-time chat content input by the first user through the instant messaging application is obtained in real time, and a corresponding candidate expression is obtained from the input candidate pool according to the real-time chat content for the first user to select.
The expression input method according to claim 1, wherein the analyzing the conversation record to obtain the emotional tendency information of the first user comprises:

Generating corresponding text data according to the session record;

Performing semantic analysis on the text data to obtain scene features and/or emotional features of the current session;

Determining the emotional tendency information of the first user according to the scene feature and/or the emotional feature of the current session.
The expression input method according to claim 2, wherein the semantic analysis of the text data to obtain scene features and/or emotion features of the current session comprises:

Performing semantic analysis on the conversation subject content of the text data to obtain scene features of the current session, wherein different scene features have different scene key values; and/or,

Performing semantic analysis on the chat content input by the first user in the text data to obtain an emotional feature of the first user, wherein different emotional features have different emotional feature values.
The expression input method according to claim 2, wherein the generating the corresponding text data according to the session record comprises:

Converting an emoticon in the session record to a corresponding tag text when an emoticon exists in the session record;

The text data is generated based on the converted session record.
The expression input method according to claim 4, wherein the generating the corresponding text data according to the session record further comprises:

When the voice content exists in the session record, the voice content is identified to obtain a corresponding text;

Translating the voice content in the session record into corresponding text, and generating corresponding text data of the session record.
The expression input method according to claim 1, further comprising: assigning an expression in the expression library before placing an expression matching the emotion preference information of the first user in the expression library into the input candidate pool Individual scene key values and/or emotional feature values.
The expression input method according to claim 1, wherein the real-time chat content input by the first user through the instant messaging application is acquired in real time, and is matched from the input candidate pool according to the real-time chat content. The steps of the corresponding candidate expression for the first user selection include:

Acquiring keywords in the real-time chat content input by the first user through the instant messaging application in real time;

Calculating a similarity between the keyword and a tag text or an index word of an expression in the input candidate pool;

Matching an expression of the input candidate pool with the highest similarity to the keyword for selection by the first user; or

Acquiring keywords in the real-time chat content input by the first user through the instant messaging application in real time;

Determining a contextual scene and/or an emotional tendency based on keywords in the live chat content;

Calculating a similarity between the contextual scene and/or sentiment orientation and the tag text or index term of the expression in the input candidate pool;

Matching an expression in the input candidate pool that is the most similar to the context scenario for the first user to select.
An expression input device comprising:

An obtaining module, configured to obtain a session record of the instant messaging application, where the chat content of the first user and other users is recorded in the session record;

An analysis module, configured to analyze the session record to obtain sentiment orientation information of the first user;

a first processing module, configured to place an expression in the expression library that matches the sentiment orientation information of the first user into the input candidate pool;

a second processing module, configured to acquire real-time chat content input by the first user through the instant messaging application, and obtain corresponding candidate expressions from the input candidate pool according to the real-time chat content for the first A user chooses.
The expression input device of claim 8, wherein the analysis module comprises:

Generating unit, configured to generate corresponding text data according to the session record;

An analyzing unit, configured to perform semantic analysis on the text data to obtain a scene feature and/or an emotional feature of the current session;

The first processing unit is configured to determine the sentiment orientation information of the first user according to the scene feature and/or the emotional feature of the current session.
The expression input device according to claim 9, wherein the analysis unit comprises:

a first analysis subunit, configured to perform semantic analysis on the session content of the text data to obtain a scene feature of the current session, wherein different scene features have different scene key values;

The second analysis subunit is configured to perform semantic analysis on the chat content input by the first user in the text data to obtain an emotional feature of the first user, wherein different emotional features have different emotional feature values.
The expression input device according to claim 9, wherein the generating unit comprises:

a first conversion subunit, configured to convert an emoticogram in the session record into a corresponding label text when an emoticon exists in the session record;

The first generation subunit is configured to generate the text data according to the converted session record.
The expression input device according to claim 9, wherein the generating unit further comprises:

a second conversion subunit, configured to: when the voice content exists in the session record, identify the voice content to obtain a corresponding text;

The second generation subunit is configured to convert the voice content in the session record into corresponding text, and generate corresponding text data of the session record.
The expression input device of claim 8, the device further comprising:

The pre-processing module is configured to: after the first processing module puts the expression matching the sentiment orientation information of the first user in the expression library into the input candidate pool, assign the respective scene key values to the expressions in the expression library and/or Or emotional feature values.
The expression input device of claim 8, wherein the second processing module comprises:

a first acquiring unit, configured to acquire, in real time, a keyword in the real-time chat content input by the first user through the instant messaging application;

a first calculating unit, configured to calculate a similarity between the keyword and a label text or an index word of an expression in the input candidate pool;

a first matching unit, configured to match an expression that is the most similar to the keyword in the input candidate pool for the first user to select;

a second acquiring unit, configured to acquire, in real time, a keyword in the real-time chat content input by the first user through the instant messaging application;

a second processing unit configured to determine a contextual scene and/or an emotional tendency based on keywords in the live chat content

a second calculating unit, configured to calculate a similarity between the context scene and/or the sentiment orientation and the label text or the index word of the expression in the input candidate pool;

And a second matching unit, configured to match an expression in the input candidate pool that is the most similar to the context scenario for the first user to select.
A terminal comprising the expression input device according to any one of claims 8 to 14.
A computer readable storage medium storing computer executable instructions that, when executed by a processor, implement the expression input method of any one of claims 1 to 7.