Evaluation processing method, device, computer system and medium

Abstract

The present disclosure provides a comment processing method, comprising: preprocessing received comments to obtain attribute words; determining evaluation words based on the distance between the attribute words and adjacent adjectives to obtain an evaluation unit, wherein the evaluation unit includes at least a set of attribute words and evaluation words; and obtaining the sentiment of the comment based at least on the evaluation words in the evaluation unit. The present disclosure also provides a comment processing device, a computer system, and a computer-readable storage medium.

Description

Technical Field

[0001] The present disclosure relates to the field of Internet technology, and more specifically, to an evaluation processing method, device, computer system, and medium. Background Art

[0002] The internet has long been a part of our lives, and nearly every online platform offers user review functionality. Authentic and valuable user reviews can significantly help other users understand the subject being reviewed. The subject being reviewed can be a specific item or service.

[0003] In the process of implementing the concepts of this disclosure, the inventors discovered that existing technologies present at least the following issues: not all users benefit from comments. Analysis reveals that this is due to the large number of internet users, the large number of comments posted, and the large number of words in these comments. Furthermore, users are accustomed to fragmented reading, making it difficult for them to discover and understand the core ideas expressed in comments within the time available. Therefore, specific methods are needed to uncover the core and important ideas in comments so that valuable information can be more effectively presented to users. Summary of the Invention

[0004] In view of this, the present disclosure provides an evaluation processing method, device, computer system and medium that facilitates extracting core important ideas from comments, so that valuable information in the evaluation can be more effectively presented to users.

[0005] One aspect of the present disclosure provides a review processing method, comprising: first, preprocessing received reviews to obtain attribute words; then, determining evaluation words based on the distance between the attribute words and adjacent adjectives to obtain an evaluation unit, wherein the evaluation unit includes at least a set of attribute words and evaluation words; and then, obtaining the sentiment of the review based at least on the evaluation words of the evaluation unit. Because the evaluation words are determined based on the distance between the attribute words and adjacent adjectives, it is possible to identify evaluation words that are not directly adjacent to the attribute words, thereby improving the accuracy of the evaluation word recognition and, in turn, improving the accuracy of the determined sentiment, making it easier for users to directly obtain the sentiment of the review.

[0006] According to an embodiment of the present disclosure, the preprocessing of the received comments to obtain attribute words may include the following operations: first, the comments are segmented based on at least a first word segmentation tool to obtain words and corresponding parts of speech, wherein the parts of speech include nouns, and the first word segmentation tool is a word segmentation tool that is based on the FudanNLP word segmentation tool and is secondary developed at the source code level to expand the supplementary dictionary function. The supplementary dictionary includes commonly used terms in the comments, and then, the words with the part of speech being nouns are matched in the attribute vocabulary to obtain attribute words.

[0007] According to an embodiment of the present disclosure, the attribute word library is an attribute word cluster library, each cluster has a baseline attribute word or a baseline attribute word and at least one similar attribute word, and the attribute word is obtained by matching words with a noun part of speech in the attribute word library, which may include obtaining the attribute word by matching words with a noun part of speech in the attribute word cluster library.

[0008] According to an embodiment of the present disclosure, the parts of speech include nouns, negative adverbs and adjectives. For a comment in the received comments, the evaluation word is determined based on the distance between the attribute word and the adjacent adjective, and the evaluation unit may include the following operations: first, the words of the comment are traversed, and the first attribute word is used as a candidate attribute word. Then, a specified number of words after the candidate attribute word is traversed, wherein, if another attribute word is traversed before the adjective is traversed, the other attribute word is used as a candidate attribute word. If the semantic edge is traversed before the adjective is traversed, the semantic edge is used as a candidate attribute word. Boundary, take the first attribute word after the semantic boundary as the candidate attribute word, repeatedly traverse the specified number of words after the candidate attribute word, until the adjective is traversed among the specified number of words after the candidate attribute word, take the candidate attribute word as the attribute word, and take the adjective as the evaluation word, then, if no negative adverb is included between the attribute word and the evaluation word, the evaluation unit includes the attribute word and the evaluation word, and then, if a negative adverb is included between the attribute word and the evaluation word, the evaluation unit includes the attribute word, the negative adverb and the evaluation word.

[0009] According to an embodiment of the present disclosure, obtaining the emotional tendency of the review based at least on the evaluation words of the evaluation unit may include the following operations: first, matching the evaluation unit in an attribute emotion library to obtain the emotional tendency of the evaluation unit, the emotional tendency including: positive, negative and neutral, wherein the attribute emotion library includes at least one group of attribute words, evaluation words and emotional tendency identifiers, and / or the attribute emotion library includes at least one group of attribute words, negative adverbs, evaluation words and emotional tendency identifiers, and then, the emotional tendency of the evaluation unit is used as the emotional tendency of the review.

[0010] According to an embodiment of the present disclosure, the method may further include the following operations: first, if the evaluation unit has no matching results in the attribute emotion library, the evaluation unit is sent, and then, in response to receiving a first instruction, the emotional tendency of the evaluation unit is determined according to the first instruction, the first instruction includes emotional tendency information, and then, the attribute emotion library is updated using the evaluation unit without a matching result and the corresponding emotional tendency.

[0011] According to an embodiment of the present disclosure, obtaining the sentiment tendency of the review based at least on the evaluation words of the evaluation unit may include the following operations: first, matching the adjectives in the evaluation unit in the sentiment word library to obtain the sentiment tendency of the adjectives in the evaluation unit, the sentiment tendency including: positive, negative and neutral; then, if the evaluation unit includes negative adverbs, the sentiment tendency of the evaluation unit is opposite to the sentiment tendency of the adjectives in the evaluation unit; if the evaluation unit does not include negative adverbs, the sentiment tendency of the evaluation unit is the same as the sentiment tendency of the adjectives in the evaluation unit; then, the sentiment tendency of the evaluation unit is used as the sentiment tendency of the review.

[0012] According to an embodiment of the present disclosure, the method may further include the following operations: after obtaining the emotional tendency of the comment, sending the emotional tendency of the comment for display on the terminal. Specifically, it may include the following operations: first, obtaining the attribute words of the evaluation unit of at least one comment, and then sorting them in descending order according to the number of the attribute words to obtain a specified number of attribute words, and then respectively obtaining the number of positive emotional tendencies and the number of negative emotional tendencies corresponding to each attribute word in the specified number of attribute words, and then sending the specified number of attribute words and the corresponding number of positive emotional tendencies and the number of negative emotional tendencies for visual display on the terminal.

[0013] According to an embodiment of the present disclosure, the method may further include the following operations: first, if a second instruction is received after the emotional tendency of the comment is sent, the second instruction includes attribute word information and emotional tendency information, and then, in response to receiving the evaluation word, at least one comment corresponding to the attribute word and emotional tendency in the attribute word information and emotional tendency information is sent for display by the terminal, wherein the evaluation unit of the at least one comment is displayed in a manner different from that of the remaining words of the at least one comment.

[0014] According to an embodiment of the present disclosure, the part of speech includes stop words, and the method may further include the following operation: after obtaining the words and the corresponding part of speech, filtering the stop words.

[0015] Another aspect of the present disclosure provides a comment processing device, including an attribute word acquisition module, an evaluation unit acquisition module and a sentiment tendency acquisition module, wherein the attribute word acquisition module is used to preprocess the received comments to obtain attribute words, the evaluation unit acquisition module is used to determine the evaluation words based on the distance between the attribute words and adjacent adjectives to obtain the evaluation units, the evaluation units include at least a group of attribute words and evaluation words, and the sentiment tendency acquisition module is used to obtain the sentiment tendency of the comments based at least on the evaluation words of the evaluation units.

[0016] According to an embodiment of the present disclosure, the attribute word acquisition module may include a word segmentation unit and an attribute word acquisition unit, wherein the word segmentation unit is used to segment the comment based on at least a first word segmentation tool to obtain words and corresponding parts of speech, wherein the parts of speech include nouns, and the first word segmentation tool is a word segmentation tool that is based on the FudanNLP word segmentation tool and is secondary developed at the source code level to expand the supplementary dictionary function, and the supplementary dictionary includes commonly used terms in the comments, and the attribute word acquisition unit is used to use words with the part of speech being nouns to match in the attribute vocabulary to obtain attribute words.

[0017] According to an embodiment of the present disclosure, the attribute word library is an attribute word cluster library, each cluster has a baseline attribute word or a baseline attribute word and at least one similar attribute word, and the attribute word acquisition unit is specifically used to use words with noun parts of speech to match and obtain attribute words in the attribute word cluster library.

[0018] According to an embodiment of the present disclosure, the parts of speech include nouns, negative adverbs and adjectives. For a comment among the received comments, the evaluation unit acquisition module may include: a selected attribute word acquisition unit, a traversal unit, an evaluation word acquisition unit, a first evaluation unit acquisition unit and a second evaluation unit acquisition unit. Among them, the attribute word acquisition unit for selection is used to traverse the words of the comment, and take the first attribute word as the attribute word for selection. The traversal unit is used to traverse a specified number of words after the attribute word for selection, wherein, if another attribute word is traversed before traversing to an adjective, the other attribute word is used as the attribute word for selection; if the semantic boundary is traversed before traversing to an adjective, the first attribute word after the semantic boundary is used as the attribute word for selection; the evaluation word acquisition unit is used to repeatedly traverse a specified number of words after the attribute word for selection until an adjective is traversed among the specified number of words after the attribute word for selection, take the attribute word for selection as the attribute word, and take the adjective as the evaluation word; the first evaluation unit acquisition unit is used to, if no negative adverb is included between the attribute word and the evaluation word, then the evaluation unit includes the attribute word and the evaluation word; the second evaluation unit acquisition unit is used to, if a negative adverb is included between the attribute word and the evaluation word, then the evaluation unit includes the attribute word, the negative adverb, and the evaluation word.

[0019] According to an embodiment of the present disclosure, the sentiment tendency acquisition module may include: a first matching unit and a first sentiment tendency acquisition unit, wherein the first matching unit is used to match the evaluation unit in the attribute sentiment library to obtain the sentiment tendency of the evaluation unit, and the sentiment tendency includes: positive, negative and neutral, wherein the attribute sentiment library includes at least one group of attribute words, evaluation words and sentiment tendency identifiers, and / or the attribute sentiment library includes at least one group of attribute words, negative adverbs, evaluation words and sentiment tendency identifiers, and the first sentiment tendency acquisition unit is used to use the sentiment tendency of the evaluation unit as the sentiment tendency of the comment.

[0020] According to an embodiment of the present disclosure, the device may further include a first sending module, a first receiving module and an updating module, wherein the first sending module is used to send the evaluation unit if the evaluation unit has no matching results in the attribute emotion library, the first receiving module is used to determine the emotional tendency of the evaluation unit according to the first instruction in response to receiving a first instruction, the first instruction includes emotional tendency information, and the updating module is used to update the attribute emotion library using the evaluation unit without matching results and the corresponding emotional tendency.

[0021] According to an embodiment of the present disclosure, the sentiment tendency acquisition module may include: a second matching unit, a second sentiment tendency acquisition unit and a third sentiment tendency acquisition unit, wherein the second matching unit is used to match the adjectives in the evaluation unit in the sentiment vocabulary to obtain the sentiment tendency of the adjectives in the evaluation unit, and the sentiment tendency includes: positive, negative and neutral; the second sentiment tendency acquisition unit is used to: if the evaluation unit includes a negative adverb, then the sentiment tendency of the evaluation unit is opposite to the sentiment tendency of the adjective in the evaluation unit; if the evaluation unit does not include a negative adverb, then the sentiment tendency of the evaluation unit is the same as the sentiment tendency of the adjective in the evaluation unit; the third sentiment tendency acquisition unit is used to use the sentiment tendency of the evaluation unit as the sentiment tendency of the comment.

[0022] According to an embodiment of the present disclosure, the device may further include a second sending module, which is used to send the emotional tendency of the comment after obtaining the emotional tendency of the comment for display on the terminal. Specifically, the second sending module may include: a first acquisition unit, a sorting unit, a second acquisition unit and a sending unit, wherein the first acquisition unit is used to obtain the attribute words of the evaluation unit of at least one comment, the sorting unit is used to sort in descending order according to the number of the attribute words to obtain a specified number of attribute words, the second acquisition unit is used to respectively obtain the number of positive emotional tendencies and the number of negative emotional tendencies corresponding to each attribute word in the specified number of attribute words, and the sending unit is used to send the specified number of attribute words and the corresponding number of positive emotional tendencies and the number of negative emotional tendencies for display on the terminal.

[0023] According to an embodiment of the present disclosure, the device may further include: a second receiving module and a third sending module, wherein the second receiving module is used to receive a second instruction after sending the emotional tendency of the comment, and the second instruction includes attribute word information and emotional tendency information; the third sending module is used to send at least one comment corresponding to the attribute word and emotional tendency in the attribute word information and emotional tendency information in response to receiving the evaluation word for display at the terminal, wherein the evaluation unit of the at least one comment is displayed in a different manner from the display method of the remaining words of the at least one comment.

[0024] According to an embodiment of the present disclosure, the part of speech includes stop words, and the apparatus may further include a filtering module, wherein the filtering module is configured to filter the stop words after obtaining the words and the corresponding part of speech.

[0025] Another aspect of the present disclosure provides a computer system, including: one or more processors, and a storage device, wherein the storage device is configured to store executable instructions, and when the executable instructions are executed by the processors, the method described above is implemented.

[0026] Another aspect of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the method described above when executed.

[0027] Another aspect of the present disclosure provides a computer program, which includes computer-executable instructions. When the instructions are executed, the computer program is used to implement the method described above.

[0028] According to the embodiments of the present disclosure, it is possible to at least partially solve the problem that users have become accustomed to fragmented reading and it is difficult for users to discover and understand the core ideas expressed in the comments within the time allowed, and thus achieve the technical effect of facilitating more effective presentation of valuable information to users.

[0029] According to the embodiments of the present disclosure, existing technologies often only recognize evaluation words that immediately follow attribute words. Recognition is ineffective when other words are between the attribute words and the evaluation words. For example, in the sentence "The price is more expensive than expected," it is difficult to accurately recognize the core idea "the price is expensive." The technical solution of the present disclosure can at least partially solve this problem and help improve the accuracy of evaluation word recognition.

[0030] According to the embodiments of the present disclosure, the existing technology relies heavily on manual labor to define review objects (attribute words), and manual methods are difficult to exhaustively identify all attribute words. For example, attributes include color and size. On the other hand, Chinese has a rich variety of synonyms. In addition to "price" itself, words describing price also include "price," "price," and "cost-effectiveness." The solution provided by the present disclosure effectively improves the coverage of the attribute word library through clustering, which helps improve the accuracy of attribute word recognition.

[0031] According to the embodiments of the present disclosure, the prior art often ignores the role of modifiers preceding evaluation words, especially negative modifiers that affect sentiment, which can lead to inconsistent or even opposite sentiment recognition results. The solution provided by the present disclosure effectively improves the accuracy of determining sentiment through the recognition of negative adverbs.

[0032] According to an embodiment of the present disclosure, for a label like "the price is high", the evaluation word "high" itself has no obvious sentiment tendency, so there will be no positive or negative return results when matching with the sentiment dictionary, but it is obvious that for the attribute "price", "high" represents a negative emotion. In this case, it is impossible to judge the sentiment tendency of the label "the price is high" simply based on the sentiment dictionary matching method. The solution provided by the present disclosure can directly use the constructed attribute sentiment library for matching, wherein the attribute sentiment library includes at least one group of attribute words, evaluation words and sentiment tendency identifiers, and / or the attribute sentiment library includes at least one group of attribute words, negative adverbs, evaluation words and sentiment tendency identifiers. By introducing the attribute-level sentiment dictionary, the sentiment tendency recognition accuracy of the evaluation words is improved.

[0033] According to the embodiments of the present disclosure, existing technologies often simply list attribute-evaluation word pairs after extracting them, without intuitively distinguishing the emotional tendencies of the evaluations, that is, the application of positive and negative evaluations, requiring users to judge the evaluation tendency themselves. The method provided by the present disclosure performs multi-attribute dimensional analysis on evaluation tags and can send the analysis results to the terminal for multi-dimensional visual display, thereby improving the application value of the tags and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0035] Figure 1 Schematically illustrates the application scenarios of the evaluation processing method, device, computer system, and medium according to the embodiments of the present disclosure;

[0036] Figure 2A The following schematically shows a flow chart of an evaluation processing method according to an embodiment of the present disclosure;

[0037] Figure 2B Schematically shows a schematic diagram of attribute clustering effect in a two-dimensional space vector according to an embodiment of the present disclosure;

[0038] Figure 3A A flowchart schematically illustrates an evaluation processing method according to another embodiment of the present disclosure;

[0039] Figure 3B The following schematically shows the effect of visual display according to an embodiment of the present disclosure;

[0040] Figure 3C Schematically shows a schematic diagram of the effect of visual display according to another embodiment of the present disclosure;

[0041] Figure 4 A block diagram schematically illustrates an evaluation processing apparatus according to an embodiment of the present disclosure; and

[0042] Figure 5 The block diagram schematically shows a computer system suitable for implementing the evaluation processing method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0043] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present disclosure. In the detailed description below, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0044] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise," "include," etc. used herein indicate the presence of the features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0045] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0046] When expressions such as “at least one of A, B, and C, etc.” are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (e.g., “a system having at least one of A, B, and C” should include but is not limited to systems having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, and C, etc.). When expressions such as “at least one of A, B, or C, etc.” are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (e.g., “a system having at least one of A, B, or C” should include but is not limited to systems having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, and C, etc.). Those skilled in the art should also understand that any transitional conjunctions and / or phrases that essentially represent two or more optional items, whether in the specification, claims, or drawings, should be understood to provide the possibility of including one, either, or both of these items. For example, the phrase "A or B" should be understood to include the possibilities of "A" or "B", or "A and B".

[0047] The process commonly used in existing technologies for mining the core ideas of reviews is primarily manual with technical assistance. For example, first, the attribute words of the review object are defined. Then, through methods such as keyword / keyword matching, these attributes and the evaluation words that describe them are found. For example, if the review content is "This shirt is not expensive and the packaging is exquisite," the "expensive" in "not expensive" and the "exquisite" in "exquisite packaging" can be identified. Next, the identified evaluation words are matched with the sentiment dictionary. If the match result is positive, the label is determined to be a positive evaluation of this attribute; conversely, if the match result is negative, the label is determined to be a negative evaluation of this attribute. In the above example, "expensive" would be identified as a negative evaluation, while "exquisite" would be identified as a positive evaluation.

[0048] However, the existing technology has the following shortcomings: first, it relies heavily on manual labor to define the review object, and manual methods are difficult to exhaust all attribute words, such as color, size, etc.; on the other hand, Chinese has a rich variety of synonyms. In addition to "price" itself, words that describe price also include variants such as "price", "price range", and "price-performance ratio", which will cause some attribute words to be omitted, and in turn cause the evaluation words corresponding to the attribute words to be omitted.

[0049] Second, existing methods can only identify evaluation words that appear immediately after attribute words. When there are other words between the attribute words and the evaluation words, recognition is ineffective. For example, in the sentence "The price is more expensive than expected," it is difficult to effectively identify the core idea "the price is expensive."

[0050] Thirdly, existing methods often ignore the role of modifiers before the evaluation words, especially negative modifiers that have an impact on emotional tendencies. This will cause the emotional tendency recognition results to be inconsistent with reality, or even opposite. Giving opposite emotional tendency results is a phenomenon that needs to be avoided as much as possible.

[0051] Furthermore, for a tag like "price is high," the evaluation word "high" itself has no obvious sentiment, so matching it with the sentiment dictionary won't return a positive or negative result. However, for the attribute "price," "high" clearly indicates a negative sentiment. In this case, simply matching with the sentiment dictionary is unable to determine the sentiment of the tag "price is high," resulting in low accuracy in the sentiment predictions generated by existing technologies.

[0052] In addition, after extracting attribute-evaluation word pairs, existing technologies often simply list them, and rarely intuitively distinguish the emotional tendencies of the evaluations, that is, the application of positive and negative evaluations. Users need to judge the evaluation tendencies themselves, which is not intuitive enough.

[0053] The embodiments of the present disclosure provide an evaluation processing method, device, computer system, and medium that facilitates the more effective presentation of valuable information in comments to users. Through operations such as clustering, word distance judgment, attribute-based sentiment dictionary, and sentiment tendency visualization, the core ideas in lengthy user comments can be extracted more efficiently, accurately, and intuitively for easy presentation to users. In one embodiment, the method includes an evaluation unit acquisition process and a sentiment tendency determination process. In the evaluation unit acquisition process, the received comments are preprocessed to obtain attribute words. Then, based on the distance between the attribute words and adjacent adjectives, evaluation words are determined to obtain evaluation units, which include at least a group of attribute words and evaluation words. After obtaining the evaluation unit, the sentiment tendency determination process is entered to obtain the sentiment tendency of the comment based on at least the evaluation words of the evaluation unit. Since the evaluation words are determined based on the distance between the attribute words and adjacent adjectives, it is possible to identify evaluation words that are not directly adjacent to the attribute words, improve the accuracy of the evaluation word recognition, and thus improve the accuracy of the determined sentiment tendency, so that users can directly obtain the sentiment tendency of the comment.

[0054] Figure 1 The application scenarios of the evaluation processing method, device, computer system and medium according to the embodiments of the present disclosure are schematically shown. Figure 1 The exemplary system architecture 100 is shown. Note that, Figure 1 The examples shown are merely examples of system architectures to which the embodiments of the present disclosure may be applied, to help those skilled in the art understand the technical content of the present disclosure, but do not mean that the embodiments of the present disclosure may not be used in other devices, systems, environments or scenarios.

[0055] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium for providing a communication link between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0056] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications with evaluation functions and / or evaluation display functions can be installed on terminal devices 101, 102, and 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0057] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0058] The server 105 may be a server that provides various services, such as a background management server (for example only) that supports websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process received user requests and other data (such as evaluation data requests) and feed back the processing results (such as web pages, information, or data obtained or generated according to the user requests) to the terminal device.

[0059] It should be noted that the evaluation processing method provided in the embodiments of the present disclosure can generally be executed by the server 105. Accordingly, the evaluation processing apparatus provided in the embodiments of the present disclosure can generally be set in the server 105. The evaluation processing method provided in the embodiments of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the evaluation processing apparatus provided in the embodiments of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0060] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0061] Figure 2A The flowchart of the evaluation processing method according to the embodiment of the present disclosure is schematically shown.

[0062] like Figure 2A As shown, the method may include operations S201 to S203.

[0063] In operation S201 , the received comments are pre-processed to obtain attribute words.

[0064] In this embodiment, the comments can be segmented, and then the attribute words in the comments can be obtained based on the segmentation results (such as the words obtained after segmentation and the corresponding parts of speech), where the attribute words are usually nouns. Therefore, the nouns can be used to match in a pre-built attribute dictionary to determine whether the noun is an attribute word. The attribute words can be manually defined comment object attribute words, such as price, appearance, packaging, logistics, etc.

[0065] Specifically, word segmentation primarily refers to separating words from a sentence, specifically, breaking a sequence of Chinese characters into individual words. Part-of-speech tagging, also known as part-of-speech tagging or simply tagging, is the process of assigning an appropriate part of speech to each word in a sentence, determining whether it is a noun, verb, adjective, or something else entirely. For example, the original review reads: "This shirt is not expensive." Accordingly, after word segmentation and part-of-speech tagging, the result is: this\pronoun piece\quantifier shirt\noun price\noun not\adverb expensive\adjective.

[0066] In one embodiment, preprocessing the received comments to obtain attribute words may include the following operations: first, segmenting the comments based on at least a first word segmentation tool to obtain words and corresponding parts of speech, where the parts of speech include nouns. The first word segmentation tool is a word segmentation tool that is developed at the source code level based on the FudanNLP word segmentation tool to extend the supplementary dictionary function. The supplementary dictionary includes common terms in the comments. Then, using words with the part of speech of nouns to match in the attribute vocabulary to obtain attribute words.

[0067] It should be noted that there are many word segmentation and part-of-speech tagging tools at present, and the ictclas word segmentation tool of the Chinese Academy of Sciences is generally used by default. However, the present disclosure does not follow the authority, but makes a comprehensive comparison of multiple word segmentation tools in terms of accuracy and performance, and finally finds that the FudanNLP word segmentation tool is most suitable for word segmentation and part-of-speech tagging of commentary corpus. However, after deciding to use the Fudan NLP word segmentation tool, the original version of the Fudan NLP word segmentation tool does not support the dictionary function. Since specific words appear in the comments, such as commonly used terms in the comments, it is very necessary to supplement the dictionary-assisted word segmentation function. Therefore, based on the FudanNLP word segmentation tool, secondary development at the source code level is carried out to expand the word segmentation tool with the supplementary dictionary function, which realizes the dictionary-assisted word segmentation function and further improves the accuracy of the word segmentation of the comments.

[0068] In addition, regarding the acquisition of attribute words, the conventional practice is usually to match the nouns in the comments (attribute words are almost all nouns) with the manually provided attribute word library, and then identify the corresponding evaluation words based on the matched attribute results. This means that if a word does not appear in the attribute word library, it will not be recognized. As mentioned above, there are many synonyms in Chinese. For example, in addition to "price" itself, there are also variants such as "price", "price", and "price performance" to describe the price. It is difficult for the attribute word library to exhaust all these variants, so this method of relying on the word library has the possibility of missing attribute words. To solve this problem, the present disclosure adopts a method based on text clustering to automatically cluster words with similar meanings together to obtain an attribute word cluster library, which improves the problem of missing attribute words to a certain extent.

[0069] Specifically, the attribute word library is an attribute word cluster library, each cluster has a baseline attribute word or a baseline attribute word and at least one similar attribute word, and the attribute word is obtained by matching words with a noun part of speech in the attribute word library, including obtaining attribute words by matching words with a noun part of speech in the attribute word cluster library.

[0070] Figure 2B The figure schematically shows the attribute clustering effect in a two-dimensional space vector according to an embodiment of the present disclosure.

[0071] like Figure 2B As shown, a large number of words can be collected in advance (or a large number of reviews can be collected and nouns extracted from them). The collected words can then be clustered. Each cluster can contain one base attribute word or one base attribute word and at least one similar attribute word. For example, if price is the base attribute word, similar attribute words include but are not limited to: selling price, price level, or price; if logistics is the base attribute word, similar attribute words include but are not limited to: delivery, express delivery, or distribution; if appearance is the base attribute word, similar attribute words include but are not limited to: shape, appearance, or surface; if taste is the base attribute word, similar attribute words include but are not limited to: taste, flavor, or flavor. The words in large font in the figure are base words, and the words near the base word are clustered together with similar attribute words. Specifically, the cosine distance can be used to determine the similarity of each word.

[0072] In a specific embodiment, first, a vector space model can be used to depict multiple nouns (such as multiple collected nouns or nouns in multiple collected comments) into a two-dimensional vector space. Then, the Word2Vec algorithm is used to cluster the reference words as the centroid, and words with similar meanings are automatically clustered together. This effectively expands the existing attribute word library and obtains an attribute word cluster library, greatly improving the coverage of the existing attribute word library and helping to reduce attribute word omissions. Among them, the vector space model was proposed by Salton et al. in the 1970s and successfully applied to the famous SMART text retrieval system. It simplifies the processing of text content into vector operations in the vector space, and it expresses semantic similarity in spatial similarity, which is intuitive and easy to understand. The Word2Vec algorithm is an optimization algorithm based on the vector space model. Clustering is a type of unsupervised algorithm in the field of machine learning, which is mainly used to automatically discover similar data points near a reference point.

[0073] In this way, it is possible to obtain attribute words by matching words whose part of speech is noun in the attribute word clustering library, and the obtained attribute words are more accurate and comprehensive than those in the prior art.

[0074] In another embodiment, the word types include stop words, and the method may further include the following operation: after obtaining the words and their corresponding word types, filtering the stop words.

[0075] Among them, stop words refer to certain words or phrases that are automatically filtered out before or after processing natural language data (or text) in information retrieval to save storage space and improve search efficiency. These words or phrases are called stop words. Common stop words include: "de", "le", "en", etc. Since stop words appear frequently and may interfere with the analysis results to some extent, they can be filtered out from the word segmentation and word type tagging text to be processed at the initial stage.

[0076] Specifically, due to the strong regularity of Chinese stop words, this embodiment directly filters them by matching an authoritative stop word corpus. By filtering stop words, the accuracy of the obtained attribute words and evaluation words can be effectively improved and they can be made more comprehensive.

[0077] Then, in operation S202, an evaluation word is determined based on the distance between the attribute word and the adjacent adjective, and an evaluation unit is obtained. The evaluation unit includes at least one group of attribute words and evaluation words.

[0078] In this embodiment, considering that the attribute words and the corresponding evaluation words in the comment are not necessarily directly adjacent, that is, there may be other words between the attribute word and the corresponding evaluation word. In the prior art, each comment is usually traversed, and it is judged whether the adjacent word at the position of the located attribute word is an adjective (semantically, evaluation words are all adjectives). If so, the adjective is used as the evaluation word corresponding to the located attribute word. If not, the attribute word is considered invalid. This may cause the prior art to leave many useful attribute words and evaluation words. To overcome these problems, this embodiment determines the evaluation word based on the distance between the attribute word and the adjacent adjective. This is because through big data analysis, it is found that the distance between the vast majority of attribute words and their corresponding evaluation words does not exceed a certain value. For example, the distance between more than 99% of the attribute words and their corresponding evaluation words does not exceed N words. Therefore, N can be used as the maximum distance between the attribute word and the evaluation word, where N is a positive integer.

[0079] In one embodiment, the word types include nouns, negative adverbs, and adjectives. For a comment received, the operation of determining an evaluation word based on the distance between the attribute word and the adjacent adjective and obtaining an evaluation unit may include the following operations.

[0080] First, traverse the words of the comment and take the first attribute word as the candidate attribute word; then, traverse a specified number of words after the candidate attribute word, wherein, if traversing to another attribute word before traversing to an adjective, take the other attribute word as the candidate attribute word; if traversing to a semantic boundary before traversing to an adjective, take the first attribute word after the semantic boundary as the candidate attribute word; repeat the above process of traversing a specified number of words after the candidate attribute word until traversing to an adjective among the specified number of words after the candidate attribute word, take the candidate attribute word as the attribute word, and take the adjective as the evaluation word, so that the evaluation word corresponding to the attribute word can be obtained. Subsequently, the evaluation word can be directly matched against a sentiment word library to determine its sentiment orientation. Alternatively, the evaluation word, attribute word, and negation adverb can be matched against a pre-set attribute sentiment library as a whole to determine its sentiment orientation. For example, if the attribute word and the evaluation word do not contain a negation adverb, the evaluation unit includes the attribute word and the evaluation word. Alternatively, if the attribute word and the evaluation word contain a negation adverb, the evaluation unit includes the attribute word, the negation adverb, and the evaluation word. This allows the evaluation word and / or evaluation unit to be obtained to facilitate the determination of sentiment orientation. The specified number of words can be 1, 2, 5, 8, 10, 12, 15, 20, or 30, depending on the specific usage experience. Preferably, through big data analysis, it has been found that the distance between more than 99% of attribute words and corresponding evaluation words does not exceed 10. Therefore, 10 can be used as the specified number, and the longest distance between an attribute word and an evaluation word is 10. If it exceeds 10, it indicates that there is no corresponding evaluation word for the current attribute word.

[0081] In a specific embodiment, the specific identification process is as follows: assuming that there are M comments in total, traverse from the first comment. When the m-th comment locates the attribute word mp1, search backward for evaluation words (adjectives) within 10 distances. If another attribute word mp2 is encountered before the evaluation word is found, skip mp1 and continue searching from mp2; if a semantic boundary is encountered before the evaluation word is found (such as punctuation marks, etc. can be used as semantic boundaries), skip mp1 and locate the next attribute word from behind the semantic boundary; if an adjective ma1 is first found within 10 distances, it is considered that the adjective ma1 is the evaluation word mc1 of the attribute word mp1, and continue to search whether there is a negative adverb (such as "not", "not so", etc.) between mp1 and mc1. If so, mark the negative word as mn1... and so on, and eventually form a set MPNC such as {attribute mp-negative adverb mn (if any)-evaluation word mc} k , the set MPNC k This is the evaluation unit.

[0082] In this way, the evaluation words and evaluation units can be obtained, so that the emotional tendency of the review can be obtained based on the evaluation words and / or evaluation units.

[0083] In operation S203, the sentiment tendency of the review is acquired based on at least the evaluation words of the evaluation unit.

[0084] In this embodiment, the sentiment tendency of the review can be obtained based on two schemes: the first is to obtain the sentiment tendency of the review based on the evaluation words; the second is to obtain the sentiment tendency of the review based on the evaluation unit.

[0085] In a first specific embodiment, obtaining the emotional tendency of the review based at least on the evaluation words of the evaluation unit may include the following operations: first, matching the evaluation unit in an attribute emotion library to obtain the emotional tendency of the evaluation unit, wherein the emotional tendency includes: positive, negative, and neutral, wherein the attribute emotion library includes at least one set of attribute words, evaluation words, and emotional tendency identifiers, and / or the attribute emotion library includes at least one set of attribute words, negative adverbs, evaluation words, and emotional tendency identifiers. Then, the emotional tendency of the evaluation unit is used as the emotional tendency of the review.

[0086] It should be noted that the attribute sentiment database can be a highly deterministic attribute sentiment dictionary (PED) constructed based on domain knowledge. For example, it may contain terms such as "noise (negative)," "taste (positive)," and "value for money (positive)." This allows the evaluation unit as a whole to be matched against the attribute sentiment dictionary (PED) to obtain the sentiment orientation of the evaluation unit. This has the following advantages: some evaluation terms, such as "big," do not inherently have a distinct sentiment orientation, but when associated with specific attribute terms, they exhibit a specific sentiment orientation. For example, in the prior art, the evaluation term "noise (noise)" does not inherently have a distinct sentiment orientation. Accordingly, the comment would be identified as neutral in sentiment, but the sentiment associated with "noise (noise)" is clearly negative. To accurately identify such situations, this embodiment uses the evaluation unit as a whole to perform a match against a pre-constructed attribute sentiment database. This effectively identifies situations where a single evaluation term is neutral but exhibits a distinct sentiment orientation when associated with a specific attribute term. In addition, the attribute emotion library includes at least one group of attribute words, evaluation words and emotional tendency identification, and / or the attribute emotion library includes at least one group of attribute words, negative adverbs, evaluation words and emotional tendency identification. For example, as described above, the attribute emotion library may also include "not too noisy (positive)," "not too bad taste (negative)," "not cost-effective (negative)," etc.

[0087] In the first specific embodiment, in order to improve the coverage of the attribute emotion library, the method may also include the following operations: first, if the evaluation unit has no matching results in the attribute emotion library, the evaluation unit is sent, such as to a terminal for display so that the developer can view it, and then, in response to receiving a first instruction (such as the emotional tendency information manually calibrated by the developer for the evaluation unit), the emotional tendency of the evaluation unit is determined according to the first instruction, and the first instruction includes emotional tendency information, and then, the attribute emotion library is updated using the evaluation unit without matching results and the corresponding emotional tendency. In this way, the MPNC identified as "neutral" in the existing method can be k Elements with obvious emotional tendencies in the results are further supplemented into PED, and MPNC is updated at the same time. k Since PED has strong reusability, it will continue to be enriched with continuous iteration and update. k The accuracy of the results will continue to improve and the coverage will become more comprehensive.

[0088] Compared with original reviews, these evaluation units can more intuitively reflect users' core evaluation experience.

[0089] In a second specific embodiment, the emotional tendency of the review obtained at least based on the evaluation words of the evaluation unit may include the following operations: first, the adjectives in the evaluation unit are matched in the emotional word library to obtain the emotional tendency of the adjectives in the evaluation unit, and the emotional tendency includes: positive, negative and neutral. Then, if the evaluation unit includes a negative adverb, the emotional tendency of the evaluation unit is opposite to the emotional tendency of the adjective in the evaluation unit. If the evaluation unit does not include a negative adverb, the emotional tendency of the evaluation unit is the same as the emotional tendency of the adjective in the evaluation unit. Then, the emotional tendency of the evaluation unit is used as the emotional tendency of the review. In this way, the emotional tendency of each evaluation word can be simply judged using the existing authoritative emotional word library ED, and then the emotional tendency of the review can be obtained, and the opposite emotional tendency will not be given due to the failure to identify negative adverbs as in the prior art. For example, the MPNC in each element of the above set MPNC can be k Evaluation word C in k Match the words in the authoritative emotional vocabulary ED (a dictionary containing positive and negative labels). If a match is found, determine whether there is a negative adverb N k If it does not exist, it is considered that the attribute evaluation element MPNC k The sentiment tendency (positive or negative) of the word is consistent with the corresponding word in the matching sentiment vocabulary; if there is no match, the sentiment tendency is neutral. kIf an element in the set is "price-effective" and there is a "price-effective positive" record in ED, the matching result is: "price-effective positive"; MPNC k If an element in the set is "quality is not very good" and there is a "good positive" record in ED, the matching result is: "quality is not very good negative".

[0090] The comment processing method provided by this disclosure introduces the concept of distance between the attribute word and the adjacent adjective, expanding the matching range of the evaluation word corresponding to the attribute word. Furthermore, the matching range of the attribute word can be expanded through text clustering technology, overcoming the limitation of relying solely on manually provided attribute words. Furthermore, the accuracy of identifying the sentiment tendency of the evaluation word can be improved by adding negative adverb recognition. Furthermore, the accuracy of identifying the sentiment tendency of the evaluation word can be improved by introducing an attribute-level sentiment dictionary.

[0091] Figure 3A The flowchart schematically shows an evaluation processing method according to another embodiment of the present disclosure.

[0092] like Figure 3A As shown, the evaluation processing method can further perform statistical analysis on one or more evaluations in multiple dimensions, and send the statistical analysis results to the terminal so that the user can intuitively and clearly see the multi-dimensional statistical analysis results of one or more evaluations on the terminal.

[0093] In one embodiment, the method may further include operation S301.

[0094] In operation S301, after obtaining the emotional tendency of the comment, the emotional tendency of the comment is sent. Specifically, the sending of the emotional tendency of the comment may include the following operations: first, obtaining the attribute words of the evaluation unit of at least one comment, then sorting them in descending order according to the number of the attribute words to obtain a specified number of attribute words, then respectively obtaining the number of positive emotional tendencies and the number of negative emotional tendencies corresponding to each attribute word in the specified number of attribute words, and then sending the specified number of attribute words and the corresponding number of positive emotional tendencies and the number of negative emotional tendencies for visual display on the terminal. In this way, it is possible to display the analysis and statistical results of certain comments in a personalized manner on the user terminal.

[0095] Figure 3B The following schematic diagram schematically shows the effect of visual display according to an embodiment of the present disclosure.

[0096] like Figure 3BAs shown in the figure, it is a schematic diagram of the visual display of comments on a product. Among them, in the comments on the product, its attribute words obtained through statistics include price, packaging, taste, logistics, seasoning, feel, etc. Among them, price, packaging, taste, logistics, and seasoning are ranked high and occupy the vast majority of the number of comments. Therefore, price, packaging, taste, logistics, and seasoning can be used as dimensions for statistical analysis respectively. Among them, each dimension corresponds to the number of positive emotional tendencies and the number of negative emotional tendencies. The server can send the statistical results to the user terminal, and the user terminal will display them. As Figure 3B The above figure is a bar chart that shows the number of positive and negative sentiment tendencies corresponding to each dimension. Among them, price, packaging, and taste have the most comments, and price and packaging have the highest praise. Taste and logistics have more negative comments. Users who have high requirements for taste and logistics can avoid choosing this product to avoid a bad experience. Figure 3B The figure below is a radar chart that can show the proportion of positive emotional tendencies in each dimension, such as the price and packaging of the product that are most popular among buyers.

[0097] In addition, in order to improve the convenience of user use, the method may also include the following operations: For example, if the server receives a second instruction after sending the emotional tendency of the comment to the user terminal, and the second instruction includes attribute word information and emotional tendency information, then the server, in response to receiving the evaluation word, sends at least one comment corresponding to the attribute word and emotional tendency in the attribute word information and emotional tendency information for display on the terminal. In addition, the evaluation unit of the at least one comment can be displayed in a different way from the display method of the remaining words of the at least one comment, such as if there is relevant display method information in the comment sent by the server.

[0098] Figure 3C The following schematically shows the effect of visual display according to another embodiment of the present disclosure.

[0099] like Figure 3C As shown, the display interface of the user terminal can also perform human-computer interaction. For example, the user can request the required data from the server through the human-computer interaction interface of the user terminal, such as one or more comments corresponding to the positive emotional tendency under the packaging dimension, and the comments sent can further include attribute words, evaluation words, evaluation units and other information. In addition, it can also include the display method of attribute words, evaluation words, evaluation units and other information, so that the key points in the information displayed to the user are more intuitive.

[0100] In a specific embodiment, the visual interface may include two parts: Figure 3B The above picture shows the three attribute words with the most positive and negative comments for this product. Select a label. Figure 3BThe figure below shows the original comment text corresponding to the attribute words related to the tag, and highlights the specific tag content. Figure 3C By visually displaying detailed information on more dimensional attributes through bar charts and radar charts, this system allows potential buyers to present the core ideas of reviews in a very intuitive way, thereby increasing the application value and user experience of reviews to a certain extent.

[0101] Figure 4 The following schematically shows a block diagram of an evaluation processing device according to an embodiment of the present disclosure.

[0102] like Figure 4 As shown, the comment processing device 400 includes an attribute word acquisition module 410 , an evaluation unit acquisition module 420 and a sentiment tendency acquisition module 430 .

[0103] The attribute word acquisition module 410 is used to pre-process the received comments to obtain attribute words.

[0104] The evaluation unit acquisition module 420 is used to determine the evaluation word based on the distance between the attribute word and the adjacent adjective to obtain the evaluation unit. The evaluation unit includes at least a group of attribute words and evaluation words.

[0105] The sentiment tendency acquisition module 430 is used to acquire the sentiment tendency of the review based at least on the evaluation words of the evaluation unit.

[0106] The attribute word acquisition module may include a word segmentation unit and an attribute word acquisition unit.

[0107] The word segmentation unit is used to segment the comments based on at least a first word segmentation tool to obtain words and corresponding parts of speech, wherein the parts of speech include nouns. The first word segmentation tool is a word segmentation tool that is based on the FudanNLP word segmentation tool and is secondary developed at the source code level to expand the supplementary dictionary function. The supplementary dictionary includes commonly used terms in the comments.

[0108] The attribute word acquisition unit is used to obtain attribute words by matching words whose part of speech is noun in the attribute word library.

[0109] Optionally, the attribute word library is an attribute word cluster library, each cluster has a base attribute word or a base attribute word and at least one similar attribute word, and the attribute word acquisition unit is specifically used to use words with noun part of speech to match in the attribute word cluster library to obtain attribute words.

[0110] In one embodiment, the parts of speech include nouns, negative adverbs and adjectives. For a comment among the received comments, the evaluation unit acquisition module 420 may include: a candidate attribute word acquisition unit, a traversal unit, an evaluation word acquisition unit, a first evaluation unit acquisition unit and a second evaluation unit acquisition unit.

[0111] The candidate attribute word acquisition unit is used to traverse the words of the comment and take the first attribute word as the candidate attribute word.

[0112] The traversal unit is used to traverse a specified number of words after the attribute word to be selected, wherein, if another attribute word is traversed before the adjective is traversed, the other attribute word is used as the attribute word to be selected; if the semantic boundary is traversed before the adjective is traversed, the first attribute word after the semantic boundary is used as the attribute word to be selected.

[0113] The evaluation word acquisition unit is used to repeatedly traverse a specified number of words after the candidate attribute word until an adjective is found among the specified number of words after the candidate attribute word, and use the candidate attribute word as the attribute word and the adjective as the evaluation word.

[0114] The first evaluation unit acquisition unit is configured to, if no negative adverb is included between the attribute word and the evaluation word, cause the evaluation unit to include the attribute word and the evaluation word.

[0115] The second evaluation unit acquisition unit is configured to, if a negative adverb is included between the attribute word and the evaluation word, then the evaluation unit includes the attribute word, the negative adverb and the evaluation word.

[0116] In another embodiment, the emotional tendency acquisition module 430 may include: a first matching unit and a first emotional tendency acquisition unit.

[0117] In which, the first matching unit is used to match the evaluation unit in the attribute emotion library to obtain the emotional tendency of the evaluation unit, and the emotional tendency includes: positive, negative and neutral, wherein the attribute emotion library includes at least one group of attribute words, evaluation words and emotional tendency identifiers, and / or the attribute emotion library includes at least one group of attribute words, negative adverbs, evaluation words and emotional tendency identifiers.

[0118] The first sentiment tendency acquisition unit is used to use the sentiment tendency of the evaluation unit as the sentiment tendency of the review.

[0119] In addition, in order to improve the coverage of the attribute emotion library, the device 400 may further include: a first sending module 440 , a first receiving module 450 and an updating module 460 .

[0120] The first sending module 440 is configured to send the evaluation unit if the evaluation unit has no matching result in the attribute emotion library.

[0121] The first receiving module 450 is configured to determine the emotional tendency of the evaluation unit according to the first instruction in response to receiving the first instruction, where the first instruction includes emotional tendency information.

[0122] The updating module 460 is used to update the attribute emotion library using the evaluation units without matching results and the corresponding emotion tendencies.

[0123] In another embodiment, the emotional tendency acquisition module 430 may include: a second matching unit, a second emotional tendency acquisition unit, and a third emotional tendency acquisition unit.

[0124] The second matching unit is used to match the adjectives in the evaluation unit in a sentiment word library to obtain the sentiment tendency of the adjectives in the evaluation unit, and the sentiment tendency includes: positive, negative and neutral.

[0125] The second sentiment tendency acquisition unit is used to: if the evaluation unit includes a negative adverb, the sentiment tendency of the evaluation unit is opposite to the sentiment tendency of the adjective in the evaluation unit; if the evaluation unit does not include a negative adverb, the sentiment tendency of the evaluation unit is the same as the sentiment tendency of the adjective in the evaluation unit.

[0126] The third sentiment tendency acquisition unit is used to use the sentiment tendency of the evaluation unit as the sentiment tendency of the review.

[0127] In yet another embodiment, the apparatus 400 may further include a second sending module 470 .

[0128] The second sending module 470 is used to send the emotional tendency of the comment after obtaining the emotional tendency of the comment. Specifically, the second sending module 470 may include: a first obtaining unit, a sorting unit, a second obtaining unit, and a sending unit.

[0129] The first acquisition unit is used to acquire the attribute words of the evaluation unit of at least one comment.

[0130] The sorting unit is used to sort the attribute words in descending order according to the number of the attribute words to obtain a specified number of attribute words.

[0131] The second acquisition unit is used to respectively acquire the number of positive sentiment tendencies and the number of negative sentiment tendencies corresponding to each attribute word in the specified number of attribute words.

[0132] The sending unit is used to send the specified number of attribute words and the corresponding number of positive emotional tendencies and the number of negative emotional tendencies so that the terminal can perform visual display.

[0133] In other embodiments, the apparatus 400 may further include: a second receiving module 480 and a third sending module 490 .

[0134] The second receiving module is configured to receive a second instruction after sending the emotional tendency of the comment, wherein the second instruction includes attribute word information and emotional tendency information.

[0135] The third sending module is used to send at least one comment corresponding to the attribute word and emotional tendency in the attribute word information and emotional tendency information in response to receiving the evaluation word for display on the terminal, wherein the evaluation unit of the at least one comment is displayed in a different manner from the display of the remaining words of the at least one comment.

[0136] In addition, the apparatus 400 may further include a filtering module 411 , which is configured to filter stop words after obtaining the words and the corresponding parts of speech.

[0137] According to the modules, submodules, units, and subunits of the embodiments of the present invention, any multiple or at least part of the functions of any multiple thereof can be implemented in one module. According to the modules, submodules, units, and subunits of the embodiments of the present invention, any one or more thereof can be split into multiple modules for implementation. According to the modules, submodules, units, and subunits of the embodiments of the present invention, any one or more thereof can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware of any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation modes of software, hardware, and firmware or in an appropriate combination of any of them. Alternatively, according to the modules, submodules, units, and subunits of the embodiments of the present invention, one or more thereof can be at least partially implemented as a computer program module, which can perform the corresponding function when the computer program module is run.

[0138] For example, any number of the attribute word acquisition module 410, the evaluation unit acquisition module 420, the sentiment tendency acquisition module 430, the first sending module 440, the first receiving module 450, the updating module 460, the second sending module 470, the second receiving module 480, the third sending module 490, and the filtering module 411 may be combined into a single module, or any one of these modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in a single module. According to an embodiment of the present disclosure, at least one of the attribute word acquisition module 410, the evaluation unit acquisition module 420, the sentiment tendency acquisition module 430, the first sending module 440, the first receiving module 450, the update module 460, the second sending module 470, the second receiving module 480, the third sending module 490 and the filtering module 411 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or can be implemented in any one of the three implementation methods of software, hardware and firmware, or in an appropriate combination of any of them. Alternatively, at least one of the attribute word acquisition module 410, the evaluation unit acquisition module 420, the sentiment tendency acquisition module 430, the first sending module 440, the first receiving module 450, the update module 460, the second sending module 470, the second receiving module 480, the third sending module 490 and the filtering module 411 can be at least partially implemented as a computer program module, which can execute corresponding functions when the computer program module is run.

[0139] Figure 5 The block diagram of a computer system suitable for implementing the above-described method according to an embodiment of the present disclosure is schematically shown. Figure 5 The computer system shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0140] like Figure 5As shown, the computer system 500 according to an embodiment of the present disclosure includes a processor 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage part 508 into a random access memory (RAM) 503. The processor 501 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (such as an application-specific integrated circuit (ASIC)), etc. The processor 501 may also include an onboard memory for caching purposes. The processor 501 may include a single processing unit or multiple processing units for executing different actions of the method flow according to an embodiment of the present disclosure.

[0141] Various programs and data required for the operation of the system 500 are stored in the RAM 503. The processor 501, ROM 502, and RAM 503 are connected to each other via a bus 504. The processor 501 executes the programs in the ROM 502 and / or RAM 503 to perform various operations of the method flow according to the embodiment of the present disclosure. It should be noted that the programs may also be stored in one or more memories other than the ROM 502 and RAM 503. The processor 501 may also execute the programs stored in the one or more memories to perform various operations of the method flow according to the embodiment of the present disclosure.

[0142] According to an embodiment of the present disclosure, the system 500 may further include an input / output (I / O) interface 505, which is also connected to the bus 504. The system 500 may also include one or more of the following components connected to the I / O interface 505: an input section 506 including a keyboard, a mouse, etc.; an output section 507 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and speakers; a storage section 508 including a hard disk; and a communication section 509 including a network interface card such as a LAN card or a modem. The communication section 509 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 505 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed in the drive 610 as needed, so that a computer program read therefrom can be installed into the storage section 508 as needed.

[0143] According to an embodiment of the present disclosure, the method flow according to an embodiment of the present disclosure can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 509, and / or installed from the removable medium 611. When the computer program is executed by the processor 501, the above-mentioned functions defined in the system of the embodiment of the present disclosure are executed. According to an embodiment of the present disclosure, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0144] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently and not be incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, and when executed, implements the method according to the embodiments of the present disclosure.

[0145] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 502 and / or RAM 503 described above and / or one or more memories other than ROM 502 and RAM 503.

[0146] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0147] Those skilled in the art will appreciate that various combinations and / or combinations of features described in the various embodiments and / or claims of this disclosure may be made, even if such combinations or combinations are not explicitly described in this disclosure. In particular, various combinations and / or combinations of features described in the various embodiments and / or claims of this disclosure may be made, without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0148] The embodiments of the present disclosure are described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although each embodiment has been described separately above, this does not mean that the measures in each embodiment cannot be used in combination to advantage. The scope of the present disclosure is defined by the appended claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A method for processing comments, comprising: Preprocess the received comments to obtain attribute words; Determining an evaluation word based on the distance between the attribute word and the adjacent adjective to obtain an evaluation unit includes: Traverse the words of at least one comment and take the first attribute word as the candidate attribute word; Traverse the specified number of words after the candidate attribute words, where: If another attribute word is traversed before the adjective, the other attribute word is used as the candidate attribute word. If the semantic boundary is reached before the adjective is reached, the first attribute word after the semantic boundary is taken as the candidate attribute word; Repeating the traversal of a specified number of words after the candidate attribute word until an adjective is found among the specified number of words after the candidate attribute word, taking the candidate attribute word as the attribute word and the adjective as the evaluation word to obtain an evaluation unit; wherein the evaluation unit includes at least a set of attribute words and evaluation words; and Acquiring the sentiment tendency of the review based at least on the evaluation words of the evaluation unit includes: Matching the attribute words and evaluation words of at least one group of evaluation units in an attribute emotion library, wherein the attribute emotion library includes at least one group of attribute words, evaluation words, and emotional tendency identifiers; when the attribute words and evaluation words of the at least one group of evaluation units are the same as at least one group of attribute words and evaluation words in the attribute emotion library, using the emotional tendency identifiers corresponding to the attribute words and evaluation words in the attribute emotion library as the emotional tendency of the evaluation unit; The sentiment tendency of the evaluation unit is used as the sentiment tendency of the review.

2. The method according to claim 1, wherein The received comments are pre-processed to obtain attribute words including: Segmenting the comments at least based on a first word segmentation tool to obtain words and corresponding parts of speech, wherein the parts of speech include nouns, the first word segmentation tool being a word segmentation tool that is developed at the source code level based on the FudanNLP word segmentation tool to extend a supplementary dictionary function, the supplementary dictionary including commonly used terms in the comments; and Attribute words are obtained by matching words whose part of speech is noun in the attribute word library.

3. According to the method according to claim 2, the attribute word library is an attribute word cluster library, each cluster has a baseline attribute word or a baseline attribute word and at least one similar attribute word, and the attribute word is obtained by matching the word with the part of speech being a noun in the attribute word library, including obtaining the attribute word by matching the word with the part of speech being a noun in the attribute word cluster library.

4. The method according to claim 2 or 3, wherein: The parts of speech include nouns, negative adverbs and adjectives. For a comment in the received comments, If no negative adverb is included between the attribute word and the evaluation word, the evaluation unit includes the attribute word and the evaluation word; as well as If a negative adverb is included between the attribute word and the evaluation word, the evaluation unit includes the attribute word, the negative adverb and the evaluation word.

5. The method according to claim 4, wherein The emotional tendencies include: positive, negative and neutral, The attribute emotion library includes at least one set of attribute words, evaluation words and emotional tendency identifiers. and / or The attribute emotion library includes at least one group of attribute words, negative adverbs, evaluation words and emotional tendency identifiers.

6. The method according to claim 5, further comprising: If the evaluation unit has no matching result in the attribute emotion library, sending the evaluation unit; In response to receiving a first instruction, determining the emotional tendency of the evaluation unit according to the first instruction, the first instruction including emotional tendency information; as well as The attribute emotion library is updated using the evaluation units without matching results and the corresponding emotion tendencies.

7. The method according to claim 4, wherein: The step of obtaining the sentiment tendency of the review based at least on the evaluation words of the evaluation unit includes: Matching the adjectives in the evaluation unit in a sentiment word library to obtain the sentiment tendency of the adjectives in the evaluation unit, wherein the sentiment tendency includes: positive, negative and neutral; If the evaluation unit includes a negative adverb, the sentiment tendency of the evaluation unit is opposite to the sentiment tendency of the adjective in the evaluation unit; if the evaluation unit does not include a negative adverb, the sentiment tendency of the evaluation unit is the same as the sentiment tendency of the adjective in the evaluation unit; and The sentiment tendency of the evaluation unit is used as the sentiment tendency of the review.

8. The method according to any one of claims 5 to 7, wherein: The method further comprises: After obtaining the sentiment tendency of the comment, sending the sentiment tendency of the comment includes: Obtaining attribute words of an evaluation unit of at least one comment; Sort the attribute words in descending order according to the number of the attribute words to obtain a specified number of attribute words; Respectively obtaining the number of positive sentiment tendencies and the number of negative sentiment tendencies corresponding to each attribute word in the specified number of attribute words; and The specified number of attribute words and the corresponding number of positive sentiment tendencies and the number of negative sentiment tendencies are sent so that the terminal can perform visual display.

9. The method according to claim 8, further comprising: If a second instruction is received after the emotional tendency of the comment is sent, the second instruction includes attribute word information and emotional tendency information; as well as In response to receiving the evaluation word, at least one comment corresponding to the attribute word and emotional tendency in the attribute word information and emotional tendency information is sent for display at the terminal, wherein the evaluation unit of the at least one comment is displayed in a different manner from the display method of the remaining words of the at least one comment.

10. The method according to claim 2, wherein the part of speech comprises a stop word, and the method further comprises: After obtaining the words and their corresponding parts of speech, stop words are filtered out.

11. A comment processing device, comprising: The attribute word acquisition module is used to pre-process the received comments and obtain attribute words; An evaluation unit acquisition module is used to determine an evaluation word based on the distance between the attribute word and the adjacent adjective to obtain an evaluation unit, including: a candidate attribute word acquisition unit, used to traverse the words of at least one comment and take the first attribute word as the candidate attribute word; a traversal unit, used to traverse a specified number of words after the candidate attribute word, wherein, if another attribute word is traversed before the adjective is traversed, the other attribute word is taken as the candidate attribute word; if the semantic boundary is traversed before the adjective is traversed, the first attribute word after the semantic boundary is taken as the candidate attribute word; an evaluation word acquisition unit, used to repeatedly traverse a specified number of words after the candidate attribute word until an adjective is traversed among the specified number of words after the candidate attribute word, taking the candidate attribute word as the attribute word and the adjective as the evaluation word; the evaluation unit includes at least one set of attribute words and evaluation words; and The sentiment tendency acquisition module is used to obtain the sentiment tendency of the review based on at least the evaluation words of the evaluation unit, including: a first matching unit, used to match the attribute words and evaluation words of at least one group of evaluation units in the attribute emotion library, wherein the attribute emotion library includes at least one group of attribute words, evaluation words and sentiment tendency identifiers; when the attribute words and evaluation words of at least one group of evaluation units are the same as at least one group of attribute words and evaluation words in the attribute emotion library, the sentiment tendency identifier corresponding to the attribute words and evaluation words in the attribute emotion library is used as the sentiment tendency of the evaluation unit; the first sentiment tendency acquisition unit is used to use the sentiment tendency of the evaluation unit as the sentiment tendency of the review.

12. The device according to claim 11, wherein The attribute word acquisition module includes: a word segmentation unit for segmenting the comments at least based on a first word segmentation tool to obtain words and corresponding parts of speech, wherein the parts of speech include nouns, and the first word segmentation tool is a word segmentation tool that is developed at the source code level based on the FudanNLP word segmentation tool to extend the supplementary dictionary function, and the supplementary dictionary includes commonly used terms in the comments; and The attribute word acquisition unit is used to obtain attribute words by matching words whose part of speech is noun in the attribute word library.

13. According to the device of claim 12, the attribute word library is an attribute word cluster library, each cluster has a baseline attribute word or a baseline attribute word and at least one similar attribute word, and the attribute word acquisition unit is specifically used to use words with the part of speech of noun to match in the attribute word cluster library to obtain attribute words.

14. The device according to claim 12 or 13, wherein The parts of speech include nouns, negative adverbs, and adjectives. For a comment among the received comments, the evaluation unit acquisition module further includes: a first evaluation unit acquiring unit configured to, if no negative adverb is included between the attribute word and the evaluation word, include the attribute word and the evaluation word in the evaluation unit; and The second evaluation unit acquisition unit is configured to, if a negative adverb is included between the attribute word and the evaluation word, cause the evaluation unit to include the attribute word, the negative adverb and the evaluation word.

15. The device according to claim 14, wherein The emotional tendency acquisition module includes the emotional tendencies of: positive, negative and neutral, wherein: The attribute emotion library includes at least one set of attribute words, evaluation words and emotional tendency identifiers. and / or The attribute emotion library includes at least one group of attribute words, negative adverbs, evaluation words and emotional tendency identifiers.

16. The apparatus according to claim 15, further comprising: A first sending module is configured to send the evaluation unit if the evaluation unit has no matching result in the attribute emotion library; a first receiving module configured to, in response to receiving a first instruction, determine the emotional tendency of the evaluation unit according to the first instruction, wherein the first instruction includes emotional tendency information; as well as An updating module is used to update the attribute emotion library using the evaluation units without matching results and the corresponding emotion tendencies.

17. The device according to claim 14, wherein The emotional tendency acquisition module includes: A second matching unit is used to match the adjectives in the evaluation unit with the sentiment word library to obtain the sentiment tendency of the adjectives in the evaluation unit, where the sentiment tendency includes: positive, negative and neutral; a second sentiment tendency acquiring unit, configured to, if the evaluation unit includes a negative adverb, determine that the sentiment tendency of the evaluation unit is opposite to the sentiment tendency of the adjective in the evaluation unit; and, if the evaluation unit does not include a negative adverb, determine that the sentiment tendency of the evaluation unit is the same as the sentiment tendency of the adjective in the evaluation unit; and The third sentiment tendency acquisition unit is used to use the sentiment tendency of the evaluation unit as the sentiment tendency of the comment.

18. The device according to any one of claims 15 to 17, wherein The device further comprises: The second sending module is configured to send the emotional tendency of the comment after obtaining the emotional tendency of the comment, including: A first acquiring unit, configured to acquire an attribute word of an evaluation unit of at least one review; A sorting unit, configured to sort the attribute words in descending order according to the number of the attribute words to obtain a specified number of attribute words; A second acquiring unit is configured to respectively acquire the number of positive sentiment tendencies and the number of negative sentiment tendencies corresponding to each attribute word in the specified number of attribute words; and The sending unit is used to send the specified number of attribute words and the corresponding number of positive emotional tendencies and the number of negative emotional tendencies so that the terminal can perform visual display.

19. The apparatus according to claim 18, further comprising: A second receiving module is configured to receive a second instruction after sending the emotional tendency of the comment, wherein the second instruction includes attribute word information and emotional tendency information; as well as The third sending module is used to send at least one comment corresponding to the attribute word and the emotional tendency in the attribute word information and the emotional tendency information in response to receiving the evaluation word for display on the terminal, wherein the evaluation unit of the at least one comment is displayed in a different manner from the display method of the remaining words of the at least one comment.

20. The apparatus according to claim 12, wherein the part of speech comprises a stop word, and the apparatus further comprises: The filtering module is used to filter out stop words after obtaining the words and the corresponding parts of speech.

21. A computer system comprising: one or more processors; A storage device for storing executable instructions, wherein when the executable instructions are executed by the processor, the method according to any one of claims 1 to 10 is implemented.

22. A computer-readable storage medium having executable instructions stored thereon, wherein the instructions, when executed by a processor, implement the method according to any one of claims 1 to 10.

Images