Method and device for generating comment content

Abstract

The present embodiment provides a method and apparatus for generating review content, relating to the field of computer application technology. The method comprises: determining a query sequence for an article to be reviewed; searching a pre-built article review library for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence; and generating review information for the article to be reviewed based on at least one of the pre-stored articles and corresponding review information. The present embodiment implements how a new article, without having been reviewed, can generate review information closely related to the content of the article to be reviewed based on articles in the article review library and their corresponding review content, thereby achieving the purpose of automatically generating article review content.

Description

Technical Field

[0001] The present application relates to the field of computer application technology, and more specifically, to a method and device for generating review content. Background Art

[0002] With the rapid development of the internet, people's daily lives are becoming increasingly dependent on it. For example, people often search for learning materials and information they need online, browse various news and comments published online, and leave their thoughts in the comment section when they encounter topics that interest them.

[0003] When there are many messages in the comment section of an article or topic, it is easier to stimulate the desire of viewers to comment on the article or topic. However, for a new article or topic, there are always few comments in the initial period, which makes the article or topic less popular. Therefore, a method of commenting on articles is needed that can automatically generate comment content.

[0004] The applicant has found that currently most methods are for generating summaries for articles or topics, and there is no solution for automatically generating article comment content. Summary of the Invention

[0005] The present application provides a method and apparatus for generating comment content, which can solve the problem of automatically generating article comment content. The technical solution is as follows:

[0006] In a first aspect, a method for generating article comment content is provided, the method comprising:

[0007] Determine a query sequence for the article to be reviewed, and search a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence;

[0008] Generate comment information of the article to be commented on based on at least one of the pre-stored article and the corresponding comment information.

[0009] In one possible implementation, before searching a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence, the method further includes:

[0010] Obtain pre-stored articles that match the article to be reviewed, as well as review information for the pre-stored articles, by using a corpus crawler;

[0011] The pre-stored articles and their comment information are associated and stored in the article comment library.

[0012] In one possible implementation, determining a query sequence for an article to be reviewed includes:

[0013] Extracting at least one keyword from the title and / or body content of the article to be reviewed;

[0014] Generate query sequences based on various permutations and combinations of keywords.

[0015] In one possible implementation, generating comment information of the article to be reviewed based on the corresponding comment information of the pre-stored article includes:

[0016] Sorting the comment information corresponding to the pre-stored documents according to a predetermined sorting rule;

[0017] The comment information corresponding to the pre-stored articles whose sorting meets the predetermined arrangement position condition is determined as the comment information of the article to be commented.

[0018] In one possible implementation, the comment information corresponding to the pre-stored documents is sorted according to a predetermined sorting rule, including any of the following:

[0019] Determine the relevance between the comment information corresponding to each pre-stored article and the article to be reviewed, and sort the comment information corresponding to the pre-stored documents in descending order of relevance;

[0020] Determine the popularity of the comment information corresponding to each pre-stored article, and sort the comment information corresponding to the pre-stored document in descending order of popularity.

[0021] In one possible implementation, generating comment information of the article to be reviewed based on at least one of the pre-stored articles and corresponding comment information includes at least one of the following:

[0022] Generate comment information for the article to be commented on based on the pre-stored articles and corresponding comment information according to a predetermined style transfer generative algorithm;

[0023] Rewrite the comment information of the pre-stored article to obtain the comment information of the article to be commented on;

[0024] Generate comment information for the article to be commented on based on pre-stored articles.

[0025] In a second aspect, a device for generating comment content is provided, the device comprising:

[0026] A search module is used to determine a query sequence for the article to be reviewed, and to search a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence;

[0027] The generating module is used to generate the comment information of the article to be commented on according to at least one of the pre-stored articles and the corresponding comment information.

[0028] According to a third aspect, an electronic device is provided, comprising:

[0029] one or more processors;

[0030] Memory;

[0031] One or more applications, wherein the one or more applications are stored in a memory and configured to be executed by one or more processors, and the one or more programs are configured to: execute the method for generating comment content of the first aspect described above.

[0032] In a fourth aspect, a computer-readable storage medium is provided, which, when executed by a processor, implements the method for generating comment content in the first aspect.

[0033] The beneficial effects of the technical solution provided by the embodiments of the present application are:

[0034] In the solution provided by the embodiments of this application, by determining a query sequence for the article to be reviewed, a pre-built article review database is searched for pre-stored articles and corresponding review information that match the article to be reviewed, and then the article information for the article to be reviewed is generated based on various combinations of the stored articles and corresponding review information. This method demonstrates how a new article, even without any reviews, can generate review information closely related to the content of the article to be reviewed based on the articles and their corresponding reviews in the article review database, thereby achieving the goal of automatically generating article review content. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments of the present application.

[0036] Figure 1 A flowchart of a method for generating article comment content provided in an embodiment of the present application;

[0037] Figure 2 A flowchart of a keyword search method;

[0038] Figure 3 This is a schematic diagram of the article to be reviewed;

[0039] Figure 4 A schematic diagram of the structure of a device for generating article comment content provided in an embodiment of the present application;

[0040] Figure 5 A schematic diagram of the structure of an electronic device for generating article comment content provided in an embodiment of the present application. DETAILED DESCRIPTION

[0041] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and are not to be construed as limiting the present invention.

[0042] Those skilled in the art will appreciate that, unless expressly stated otherwise, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the description of the embodiments of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, "connected" or "coupled" as used herein may include wireless connections or wireless couplings. The term "and / or" used herein includes all or any one of the units and all combinations of one or more associated listed items.

[0043] In order to make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0044] The embodiments of the present application provide a method and device for generating review content, which are intended to solve the above technical problems in the prior art.

[0045] The following specific embodiments describe in detail the technical solutions of the embodiments of the present application and how the technical solutions of the embodiments of the present application solve the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0046] In the embodiment of the present application, a method for generating review content is provided, such as Figure 1 As shown, the method includes:

[0047] Step S11: Determine the query sequence of the article to be reviewed, and search the pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence. Specifically, in this step, the method for determining the query sequence of the article to be reviewed is as follows:

[0048] Extract at least one keyword from the title and / or body of the article to be reviewed; and generate a query sequence based on multiple permutations and combinations of each keyword.

[0049] Methods for extracting keywords include:

[0050] First, the required keywords can be determined manually according to actual needs. The keywords can be one or more and can be in any form such as words, phrases, etc.

[0051] Second, first extract a certain number of meaningful words, characters, or sentences from the article content and / or title. Here, we take the extracted words as an example. When multiple words are extracted, the extracted words are used as target words, and the weight of each target word is calculated. The weight calculation method can be determined by calculating the page appearance frequency or the set appearance frequency of the word. For example, the extracted keywords are "epidemic", "disease", and "hospitalization". Assuming that there are 200 words in the page, of which "epidemic" appears 100 times on the page, the page appearance frequency of the word "epidemic" is 50%; similarly, assuming that the page appearance frequency of "disease" is 20% and the page appearance frequency of "hospitalization" is 5%, the weight of the words with high page appearance frequency is increased, and they are sorted from large to small according to the weight calculation results. Finally, the target words with large weights or within the specified sorting order are used as keywords;

[0052] Third, use the TextRank algorithm to extract keywords. TextRank is an algorithm used for keyword extraction, such as Figure 2 The specific process is as follows:

[0053] Step S21, cutting the article to be reviewed into complete sentence units to obtain multiple sentences;

[0054] Step S22: For each sentence, perform word segmentation and part-of-speech tagging, and only retain words with specified parts of speech, such as nouns, verbs, and adjectives. At this time, multiple intermediate candidate keywords will be obtained;

[0055] Step S23: construct an intermediate candidate keyword graph G = (V, E), where V is a node set consisting of the intermediate candidate keywords generated in step S22, and E is an edge set. An edge between any two nodes is constructed using a co-occurrence relationship. A co-occurrence relationship means that the corresponding words co-occur in a window of length K, that is, at most K words co-occur.

[0056] Step S24, iterating the weight of each node according to step S23 until convergence;

[0057] Step S25: sort the node weights in reverse order to obtain the most important N words as the final candidate keywords;

[0058] Step S12: Generate comment information for the article to be reviewed based on at least one of the pre-stored articles and corresponding comment information. This step may use one of the following methods:

[0059] The first method uses a predetermined style transfer generative algorithm to generate review information for the article to be reviewed based on pre-stored articles and their corresponding review information. This method obtains keywords from the pre-stored articles and their review content, and generates the review content of the article to be reviewed based on the review style of the pre-stored article's review information.

[0060] The second method is to rewrite the review information of the pre-stored article to obtain the review information of the article to be reviewed. In this process, alternative words, sentences, and phrases similar to the keywords, sentences, and phrases in the review information are found and replaced, such as synonyms, to rewrite the review information of the article to be reviewed.

[0061] The third method is to generate comment information for the article to be reviewed based on the pre-stored article or the article to be reviewed. This process can be done in one of the following ways:

[0062] Method 1: Generate comment information according to the pre-set TextRank algorithm, and regard each sentence in the pre-stored article or the article to be commented on as a node. If two sentences are similar, it is considered that there is an undirected weighted edge between the nodes corresponding to the two sentences. The weight is the similarity. Then, the sentences with the highest weights can be used as comment information for the article to be commented on.

[0063] Method 2: First, the sentences in the pre-existing article or the article to be reviewed are segmented and analyzed to obtain the first candidate sentences. For example, the analysis shows that the image annotation is not suitable for review information. Second, the first candidate sentences are screened. For example, if the first candidate sentence most relevant to the article title can be used as the review information; or based on the position of the first candidate sentence, such as sentences at the beginning or end of a paragraph are more likely to express the most important information, the second candidate sentences can be screened out. Finally, the second candidate sentences with high redundancy, low coherence, and that do not reflect the core content of the article are removed, and the remaining sentences are used as the review information.

[0064] Method 3: Automatic writing in intelligent writing, which allows the machine to complete the article writing independently. This method inputs keywords and describes the main information contained in the keywords in a human-like way, and finally obtains the comment information of the article to be reviewed;

[0065] Step S13: sort the comment information of the article to be reviewed and select the final required comment information. The specific method of this step is as follows:

[0066] Sorting the comment information corresponding to the pre-stored documents according to a predetermined sorting rule;

[0067] The comment information corresponding to the pre-stored articles whose sorting meets the predetermined arrangement position condition is determined as the comment information of the article to be commented.

[0068] In this step, the comment information corresponding to the pre-stored documents is sorted according to a predetermined sorting rule, including any of the following:

[0069] First, determine the relevance between the comment information corresponding to each pre-stored article and the article to be reviewed, and sort the comment information corresponding to the pre-stored documents in descending order of relevance;

[0070] Second, determine the popularity of the comment information corresponding to each pre-stored article, and sort the comment information corresponding to the pre-stored document according to the order of the popularity value from high to low; specifically, the ranking can be determined based on how many levels of comments are included in the comment information of the pre-stored article. The more levels of comments included in the comment information, the higher the popularity of the comment. For example, a comment information also includes second-level and third-level comments that restore or reply to the comment information; or the comment information with a large number of likes is considered to have high popularity; conversely, the comment information with a small number of likes is considered to have low popularity;

[0071] This embodiment of the application also provides a method for creating an article review library, which is as follows:

[0072] Using a corpus crawler, we retrieve pre-existing articles that match the article to be reviewed, along with their comment information. We then store the pre-existing articles, along with their associated comment information, in the article comment database. Alternatively, we select articles with high similarity to the keywords, words, phrases, or sentences in the article to be reviewed as pre-existing articles. We can also manually select pre-existing articles.

[0073] As can be seen, the above-mentioned solution of the embodiment of the present application determines the query sequence of the article to be reviewed, searches the pre-built article review library for pre-stored articles and corresponding review information that match the article to be reviewed, and then generates the article information of the article to be reviewed based on various combinations of the stored articles and corresponding review information. It can be seen that this method realizes how, even if a new article has not yet been reviewed, it can generate review information closely related to the content of the article to be reviewed based on the articles in the article review library and their corresponding review content, thereby achieving the purpose of automatically generating article review content.

[0074] The above method can be applied to any article or video content that needs to be commented on, such as Figure 3As shown, the article currently has no comments and needs to be filled in by the browsing user in the comment area at the bottom of the article. The method in the above embodiment can be used to generate comment information closely related to the content of the article based on the articles in the article comment library and their corresponding comment content. In this way, some comments will be automatically generated in the comment area to attract other users who browse the article to leave corresponding comments.

[0075] The embodiment of the present application provides a device for generating review content, such as Figure 4 As shown, the device comprises:

[0076] A search module 41 is used to determine a query sequence of the article to be reviewed, and search a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence;

[0077] The generating module 42 is configured to generate comment information of the article to be reviewed based on at least one of the pre-stored articles and the corresponding comment information.

[0078] The device also includes:

[0079] An acquisition module 43 is configured to acquire, by means of a corpus crawler, pre-stored articles that match the article to be reviewed, and review information of the pre-stored articles;

[0080] The pre-stored articles and their comment information are associated and stored in the article comment library.

[0081] The search module 41 is specifically used to extract at least one keyword from the title and / or body content of the article to be reviewed;

[0082] Generate query sequences based on various permutations and combinations of keywords.

[0083] The device further comprises:

[0084] The sorting module 44 is specifically used to sort the comment information corresponding to the pre-stored documents according to a predetermined sorting rule;

[0085] The comment information corresponding to the pre-stored articles whose sorting meets the predetermined arrangement position condition is determined as the comment information of the article to be commented.

[0086] The sorting module 44 is specifically configured to sort the comment information corresponding to the pre-stored documents according to a predetermined sorting rule, including any of the following:

[0087] Determine the relevance between the comment information corresponding to each pre-stored article and the article to be reviewed, and sort the comment information corresponding to the pre-stored documents in descending order of relevance;

[0088] Determine the popularity of the comment information corresponding to each pre-stored article, and sort the comment information corresponding to the pre-stored document in descending order of popularity.

[0089] The generating module 42 is specifically configured to generate comment information of the article to be reviewed based on at least one of the pre-stored articles and corresponding comment information, including at least one of the following:

[0090] Generate comment information for the article to be commented on based on the pre-stored articles and corresponding comment information according to a predetermined style transfer generative algorithm;

[0091] Rewrite the comment information of the pre-stored article to obtain the comment information of the article to be commented on;

[0092] Generate comment information for the article to be commented on based on pre-stored articles.

[0093] In an embodiment of the present application, an electronic device is provided, comprising: a memory and a processor; and at least one program stored in the memory for execution by the processor. Compared to the prior art, the method can achieve the following: by determining a query sequence for an article to be reviewed, searching a pre-built article review library for pre-stored articles and corresponding review information that match the article to be reviewed, and then generating article information for the article to be reviewed based on various combinations of the stored articles and corresponding review information. This method can achieve the goal of automatically generating article review content, by generating review information closely related to the content of the article to be reviewed based on articles in the article review library and their corresponding review content, for a new article that has not yet been reviewed.

[0094] In an alternative embodiment, an electronic device is provided, such as Figure 5 As shown, Figure 5 The electronic device 4000 shown includes: a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004. It should be noted that in actual applications, the number of transceivers 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0095] The processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of the embodiments of this application. The processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0096] Bus 4002 may include a path for transmitting information between the above components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. Bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0097] The memory 4003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0098] The memory 4003 is used to store application code for executing the solution of the embodiment of the present application, and the execution is controlled by the processor 4001. The processor 4001 is used to execute the application code stored in the memory 4003 to implement the content shown in the above method embodiment.

[0099] Electronic devices include but are not limited to desktop computers, laptop computers, etc.

[0100] The embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer is run on a computer, the computer can execute the corresponding content in the aforementioned method embodiment. Compared with the prior art, by determining the query sequence of the article to be reviewed, searching for pre-stored articles and corresponding comment information that match the article to be reviewed in the pre-built article review library, and then generating the article information of the article to be reviewed based on various combinations of the stored articles and the corresponding comment information. It can be seen that this method realizes how a new article can generate comment information closely related to the content of the article to be reviewed based on the articles in the article review library and their corresponding comment content without being reviewed, thereby achieving the purpose of automatically generating article comment content.

[0101] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0102] The present invention provides a method for generating review content, which includes:

[0103] Step A: determining a query sequence for the article to be reviewed, and searching a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence;

[0104] Step B: generating comment information of the article to be reviewed based on at least one of the pre-stored articles and corresponding comment information.

[0105] Before searching a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence, the method further includes:

[0106] Step C, obtaining pre-stored articles that match the article to be reviewed and review information of the pre-stored articles by means of a corpus crawler;

[0107] The pre-stored articles and their comment information are associated and stored in the article comment library.

[0108] The query sequence for determining the article to be reviewed in step A includes:

[0109] Extracting at least one keyword from the title and / or body content of the article to be reviewed;

[0110] Generate query sequences based on various permutations and combinations of keywords.

[0111] After generating the comment information of the article to be commented on, it also includes:

[0112] Step D: sorting the comment information corresponding to the pre-stored documents according to a predetermined sorting rule;

[0113] The comment information corresponding to the pre-stored articles whose sorting meets the predetermined arrangement position condition is determined as the comment information of the article to be commented.

[0114] Sorting the comment information corresponding to the pre-stored documents according to a predetermined sorting rule in step D may include any of the following:

[0115] Determine the relevance between the comment information corresponding to each pre-stored article and the article to be reviewed, and sort the comment information corresponding to the pre-stored documents in descending order of relevance;

[0116] Determine the popularity of the comment information corresponding to each pre-stored article, and sort the comment information corresponding to the pre-stored document in descending order of popularity.

[0117] Generating the review information of the article to be reviewed based on at least one of the pre-stored articles and corresponding review information in step B includes at least one of the following:

[0118] Generate comment information for the article to be commented on based on the pre-stored articles and corresponding comment information according to a predetermined style transfer generative algorithm;

[0119] Rewrite the comment information of the pre-stored article to obtain the comment information of the article to be commented on;

[0120] Generate comment information for the article to be commented on based on pre-stored articles.

[0121] The above descriptions are only partial embodiments of the present invention. It should be pointed out that ordinary technicians in this technical field can make several improvements and modifications without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for generating review content, characterized in that: The method includes: Extracting at least one keyword from the title and / or body content of the article to be reviewed; Generate a query sequence for the article to be reviewed based on multiple permutations and combinations of each keyword; Searching a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence; Based on at least one of the pre-stored article and the corresponding comment information, the comment information of the article to be reviewed is generated in any of the following ways: Generate comment information of the article to be reviewed based on the pre-stored article and corresponding comment information according to a predetermined style transfer generative algorithm; Rewriting the comment information of the pre-stored article to obtain the comment information of the article to be reviewed; Generate comment information of the article to be commented on based on the pre-stored article.

2. The method according to claim 1, characterized in that Before searching a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence, the method further includes: Acquire pre-stored articles that match the article to be reviewed, and review information of the pre-stored articles, by means of a corpus crawler; The pre-stored articles and their comment information are associated and stored in an article comment library.

3. The method according to claim 1, characterized in that After generating the comment information of the article to be commented, the method further includes: Sorting the comment information corresponding to the pre-stored articles according to a predetermined sorting rule; The comment information corresponding to the pre-stored articles whose sorting meets the predetermined arrangement position condition is determined as the comment information of the article to be commented.

4. The method according to claim 3, characterized in that Sorting the comment information corresponding to the pre-stored articles according to a predetermined sorting rule includes any of the following: Determine the relevance between the review information corresponding to each pre-stored article and the article to be reviewed, and sort the review information corresponding to the pre-stored article in descending order of relevance; The popularity value of the comment information corresponding to each pre-stored article is determined, and the comment information corresponding to the pre-stored article is sorted in descending order of the popularity value.

5. The method according to claim 1, wherein Generating the review information of the article to be reviewed based on the pre-stored article includes at least one of the following: For each sentence in the pre-stored article, determine the similarity between the sentence and other sentences; using each sentence in the pre-stored article as a node, construct an edge between the nodes corresponding to two similar sentences, and determine the weight of the edge according to the similarity between the two sentences corresponding to the edge; Generate comment information for the article to be commented on based on the sentence corresponding to the edge with the highest weight; Analyze each sentence in the pre-stored article to obtain a first candidate sentence; Screening each of the first candidate sentences to obtain each of the second candidate sentences based on the relevance of each of the first candidate sentences to the title of the article to be reviewed or the position of each of the first candidate sentences in the pre-stored article; A sentence with low redundancy, high coherence and consistent with the core content of the pre-stored article is determined from each of the second candidate sentences as the comment information of the article to be commented on.

6. The method according to claim 1, wherein The step of extracting at least one keyword from the title and / or body of the article to be reviewed includes: The method for extracting each keyword of the at least one keyword is one of the following methods: Extract multiple target words from the title and / or body of the article to be reviewed, calculate the weight of each target word, sort the words from large to small according to the weight calculation results, and use the target words in the specified sort order as keywords.

7. A device for generating comment content, characterized in that: include: A search module, configured to extract at least one keyword from the title and / or body of the article to be reviewed; Generate a query sequence for the article to be reviewed based on multiple permutations and combinations of keywords; search a pre-built article review database for pre-stored articles and corresponding review information that match the article to be reviewed based on the query sequence; A generating module, configured to generate the comment information of the article to be reviewed based on at least one of the pre-stored article and the corresponding comment information, in any of the following ways: generating the comment information of the article to be reviewed based on the pre-stored article and the corresponding comment information according to a predetermined style transfer generative algorithm; Rewriting the comment information of the pre-stored article to obtain the comment information of the article to be reviewed; Generate comment information of the article to be commented on based on the pre-stored article.

8. The device according to claim 7, wherein The device further includes an acquisition module, which is configured to: Acquire pre-stored articles that match the article to be reviewed, and review information of the pre-stored articles, by means of a corpus crawler; The pre-stored articles and their comment information are associated and stored in an article comment library.

9. The device according to claim 7, wherein The device further includes a sorting module, which is configured to: Sorting the comment information corresponding to the pre-stored articles according to a predetermined sorting rule; The comment information corresponding to the pre-stored articles whose sorting meets the predetermined arrangement position condition is determined as the comment information of the article to be commented.

10. The device according to claim 9, wherein The sorting module is used to: Determine the relevance between the review information corresponding to each pre-stored article and the article to be reviewed, and sort the review information corresponding to the pre-stored article in descending order of relevance; The popularity value of the comment information corresponding to each pre-stored article is determined, and the comment information corresponding to the pre-stored article is sorted in descending order of the popularity value.

11. The device according to claim 7, wherein When generating the review information of the article to be reviewed based on the pre-stored article, the generating module is used to: For each sentence in the pre-stored article, determine the similarity between the sentence and other sentences; using each sentence in the pre-stored article as a node, construct an edge between the nodes corresponding to two similar sentences, and determine the weight of the edge according to the similarity between the two sentences corresponding to the edge; Generate comment information for the article to be commented on based on the sentence corresponding to the edge with the highest weight; Analyze each sentence in the pre-stored article to obtain a first candidate sentence; Screening each of the first candidate sentences to obtain each of the second candidate sentences based on the relevance of each of the first candidate sentences to the title of the article to be reviewed or the position of each of the first candidate sentences in the pre-stored article; A sentence with low redundancy, high coherence and consistent with the core content of the pre-stored article is determined from each of the second candidate sentences as the comment information of the article to be commented on.

12. The device according to claim 7, wherein The search module is used to: Extract multiple target words from the title and / or body of the article to be reviewed, calculate the weight of each target word, sort the words from large to small according to the weight calculation results, and use the target words in the specified sort order as keywords.

13. An electronic device, characterized in that: It includes: one or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to: execute the method for generating comment content according to any one of claims 1 to 6.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for generating comment content described in any one of claims 1 to 6 is implemented.

Images