CN108509417B

CN108509417B - Title generation method and device, storage medium and server

Info

Publication number: CN108509417B
Application number: CN201810228930.0A
Authority: CN
Inventors: 丁如敏; 叶方正; 赵田
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Shenzhen Yayue Technology Co ltd
Priority date: 2018-03-20
Filing date: 2018-03-20
Publication date: 2022-03-15
Anticipated expiration: 2038-03-20
Also published as: CN108509417A

Abstract

The application discloses a title generation method and equipment, a storage medium and a server, wherein the title generation method comprises the following steps: performing semantic analysis on a current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article; acquiring target articles which belong to the target category and have popularity degrees larger than a target threshold value from historically published articles; acquiring a sentence structure of the title of the target article and a hot word in the target article; and generating alternative titles for the current article according to the keywords, the sentence pattern structure and the hot words. By adopting the technical scheme, the exposure rate and the reading amount of the current article can be improved.

Description

Title generation method and device, storage medium and server

Technical Field

The present invention relates to the field of computer technologies, and in particular, to a title generation method and apparatus, a storage medium, and a server.

Background

With the increasing number of news recommendation application programs in the market, the content of news information is more and more abundant. Every day, various news recommendation applications can newly generate tens of thousands of articles, and how to quickly attract the attention of users through the titles of the articles is also a big problem.

For example, a user who writes one hand is good at writing science and technology articles, and has a lot of dry contents, but the titles of the articles are not good, so that many users are likely not to read the articles, the exposure and the click rate of the articles are reduced, and the users are not able to find many excellent articles.

Disclosure of Invention

Embodiments of the present invention provide a title generation method and apparatus, a storage medium, and a server, which can generate an alternative title that attracts a user's attention, thereby increasing an exposure rate and a reading amount of a current article.

In a first aspect, an embodiment of the present invention provides a title generating method, including:

performing semantic analysis on a current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article;

acquiring target articles which belong to the target category and have popularity degrees larger than a target threshold value from historically published articles;

acquiring a sentence structure of the title of the target article and a hot word in the target article;

and generating alternative titles for the current article according to the keywords, the sentence pattern structure and the hot words.

In one possible design, obtaining target articles in the historical published articles that belong to the target category and are popular with the target articles greater than a target threshold includes:

acquiring a first article which belongs to the target category and has a high quality degree greater than a first threshold value in articles stored in a historical library, wherein the high quality degree is used for representing the quality degree of the target article;

and acquiring a second article which belongs to the target category and has a popularity value larger than a second threshold value in the articles published by at least one social platform, wherein the popularity value is used for representing the reading number of the target article.

In one possible design, the obtaining a first article of the articles stored in the history repository that belongs to the target category and has a quality level greater than a first threshold includes:

acquiring at least one article belonging to the target category in articles stored in a historical library;

determining the goodness of each article in the at least one article according to the reading quantity of each article in the at least one article and the interaction characteristics, wherein the interaction characteristics are used for representing the interactive operation of a user on the articles;

selecting a first article from the at least one article having a goodness-of-merit greater than the first threshold.

In one possible design, the obtaining a second article, which belongs to the target category and has a popularity value greater than a second threshold, of the at least one social platform published article includes:

obtaining at least one article belonging to the target category in articles published by at least one social platform within a target duration range before the current time;

dividing the at least one article into at least one category, wherein the events aiming at the contents of the articles belonging to the same category are the same;

for each category, accumulating the popularity values of the articles belonging to the category on the at least one social platform to obtain popularity values corresponding to the category;

and selecting a target category from the at least one category, and determining the articles belonging to the target category as second articles with the heat value larger than the second threshold, wherein the heat value corresponding to the target category is larger than the second threshold.

In one possible design, the obtaining a sentence structure of the title of the target article and a hot word in the target article includes:

acquiring a title of the first article, and acquiring a sentence structure of the title;

and acquiring the hot words of which the occurrence frequency is greater than a third threshold value in the second article.

In one possible design, the sentence structure includes parts of speech of at least two participles of the title, the keyword includes at least two words, and the generating of the alternative title for the current article according to the keyword, the sentence structure and the hot word includes:

replacing words with the same semantics as the hot word in the at least two words by the hot word;

and assembling the at least two terms after the replacement processing into the alternative titles of the current article according to the parts of speech of each participle in the sentence pattern structure.

In a second aspect, an embodiment of the present invention provides a title generating apparatus, including:

the semantic analysis unit is used for carrying out semantic analysis on the current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article;

the first acquisition unit is used for acquiring target articles which belong to the target category and have the popularity degree larger than a target threshold value in the historically published articles;

the second obtaining unit is used for obtaining a sentence structure of the title of the target article and a hot word in the target article;

and the title generating unit is used for generating alternative titles for the current article according to the keywords, the sentence pattern structure and the hot words.

In one possible design, the first obtaining unit includes:

the first article acquisition subunit is configured to acquire a first article which belongs to the target category and has a high quality degree greater than a first threshold from among articles stored in a history library, where the high quality degree is used to indicate a quality degree of the target article;

the second article obtaining subunit is configured to obtain a second article that belongs to the target category and has a popularity value greater than a second threshold value among articles published by at least one social platform, where the popularity value is used to indicate a reading number of the target article.

In one possible design, the first article obtaining subunit is specifically configured to obtain at least one article belonging to the target category from among articles stored in the history repository;

In a possible design, the second article obtaining subunit is specifically configured to obtain at least one article belonging to the target category from among articles published by at least one social platform within a target duration range before the current time;

In one possible design, the second obtaining unit includes:

a sentence structure obtaining subunit, configured to obtain a title of the first article, and obtain a sentence structure of the title;

and the hot word acquisition subunit is used for acquiring the hot words of which the occurrence frequency is greater than a third threshold value in the second article.

In one possible design, the sentence structure includes parts of speech of at least two participles of the title, the keyword includes at least two words, and the title generating unit includes:

the replacing subunit is used for replacing the words with the same semantics as the hot word in the at least two words by adopting the hot word;

and the title assembling subunit is used for assembling the at least two replaced words into the alternative titles of the current article according to the parts of speech of each participle in the sentence pattern structure.

In a third aspect, an embodiment of the present invention provides a computer storage medium storing a plurality of instructions, the instructions being adapted to be loaded by a processor and to perform the following steps:

In a fourth aspect, an embodiment of the present invention provides a server, including: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the steps of:

In the embodiment of the invention, the target category to which the current article belongs and the keywords for representing the content of the current article are acquired, the target article which belongs to the target category and has the popularity degree larger than the target threshold value in the historically published articles is further acquired, and the alternative title which attracts the attention of the user is generated for the current article by combining the sentence pattern structure of the title of the target article and the hot words in the target article, so that the exposure rate and the reading capacity of the current article are improved.

Drawings

In order to illustrate embodiments of the present invention or technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.

Fig. 1 is a schematic flow chart of a title generating method according to an embodiment of the present invention;

fig. 2 is a schematic flow chart of another title generation method according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a title generation model according to an embodiment of the present invention;

FIG. 4 is a diagram illustrating a high-quality historical title model according to an embodiment of the present invention;

fig. 5 is a schematic diagram of an article content tag extraction model according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a hot spot mining model according to an embodiment of the present invention;

fig. 7 is a schematic structural diagram of a title generating device according to an embodiment of the present invention;

fig. 8 is a schematic structural diagram of a first obtaining unit according to an embodiment of the present invention;

fig. 9 is a schematic structural diagram of a second obtaining unit according to an embodiment of the present invention;

fig. 10 is a schematic structural diagram of a title generating unit according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of a server according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be described below with reference to the drawings in the embodiments of the present invention.

The title generation method of the embodiment of the invention can be applied to background servers corresponding to various news recommendation application programs or background servers corresponding to various social platforms, for example, if a user needs to write an article to be published on a news recommendation application program or a social platform, the user can edit the current article to be published first, the background server performs semantic analysis on the current article edited by the user to obtain a target category to which the current article belongs and a keyword for representing the content of the current article, the background server analyzes the historically published article to obtain the target article which belongs to the target category and has a popularity degree greater than a target threshold, and at least one alternative title is generated for the current article by combining a sentence structure and a hot word of the title of the target article for the user to select. Because the alternative titles are combined with the sentence structure and the hot words of the titles of the historical excellent articles, the alternative titles can attract the eyes of the user and improve the exposure rate and the reading number of the current articles.

The title generation method provided by the embodiment of the invention will be described in detail below with reference to fig. 1 to 6.

Referring to fig. 1, a flow chart of a title generating method according to an embodiment of the present invention is shown. As shown in fig. 1, the method of the embodiment of the present invention may include the following steps S101 to S104.

S101, performing semantic analysis on a current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article;

in one embodiment, the current article may be an article that the user has written to pre-post to a social platform or news feed application. In order to improve the exposure rate and the reading quantity of the current article, at least one alternative title is generated for the current article by adopting the title generation method of the embodiment of the application for the user to select.

The semantic analysis algorithm is adopted to perform word segmentation and semantic understanding on the current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article, wherein the target category may include, but is not limited to, a science category, a history category, a sports category, an entertainment category, and the like. It should be noted that, in order to improve the accuracy of extracting the keywords of the current article, manual labeling and verification may be performed after the keywords are extracted by using a semantic analysis algorithm, so as to obtain more accurate keywords.

As shown in fig. 5, which is a schematic diagram of an article content TAG extraction model provided in an embodiment of the present invention, as shown in the diagram, a semantic understanding is performed on an article by word segmentation, a keyword TAG is extracted, whether the extracted keyword is correct is further determined, if the extracted keyword is correct, the extracted keyword is stored, and if the extracted keyword is incorrect, a manual labeling mode may be adopted to perform labeling and verification, and the keyword is extracted again.

For example, if the current article is a work life of a software development engineer who introduces google company as AI in one day, the current article can be obtained to belong to a science and technology article through semantic analysis of the current article, and the extracted keywords may be keywords of one day, life, daily life, google and the like.

S102, acquiring target articles which belong to the target category and have the popularity degree larger than a target threshold value from the historically published articles;

in one embodiment, target articles in the historically published articles belonging to the target category and having a popularity greater than a target threshold are further obtained.

Alternatively, the historically published articles may be articles stored in a history repository, the target article with the popularity degree greater than the target threshold may be a first article with the quality degree greater than a first threshold in the articles stored in the history repository, and the quality degree may be used to indicate the quality degree of the articles. For example, the goodness of an article can be obtained according to the reading quantity of the article and the interaction characteristics, and the interaction characteristics can be used for representing the operation characteristics of the article, for example, the interaction characteristics can include negative comment percentage of the article, collection quantity of the article, praise quantity of the article, forward sharing times of the article, and the like. If the negative comment proportion of an article is smaller, the collection number is larger, the praise number is larger, and the forwarding sharing number is larger, the quality of the article is also larger.

Alternatively, the historically published articles may be articles published by at least one social platform within a recent period of time, and the target articles having a popularity greater than the target threshold may be second articles having a popularity value greater than a second threshold in the articles published by the at least one social platform within the recent period of time. It should be noted that, when article popularity values are counted, popularity values of articles for the same event may be accumulated to form a popularity value corresponding to one category, for example, when the popularity value of an article 1 for an event a on a microblog is a, the popularity value of an article 2 for an event a on a hectometic clouds is b, and the popularity value of an article 3 for an event a on Tencent news is c, the popularity value a + b + c is taken as the popularity value corresponding to the event a category. It should be noted that the heat value can be obtained by quantifying the number of readings of the article. The article 1, the article 2 and the article 3 corresponding to the event category a are the second articles.

Optionally, the historically published articles may include both the articles stored in the history repository and the articles published by at least one social platform in a recent period of time, and the target articles having a popularity greater than the target threshold may include the first article and the second article.

S103, acquiring a sentence structure of the title of the target article and hot words in the target article;

in one embodiment, a sentence structure of the title of the target article and a hot word in the target article are obtained, wherein the hot word may be a word in the target article, the occurrence frequency of which is higher than a third threshold value. The sentence structure of the title may be the part of speech of each participle of the title, for example, the sentence structure may be "adjective! The noun | turn | question ". And further, the participles with higher occurrence frequency in the title of the target article can be obtained. For example, the part of speech with question "do you dislike", the part of speech with adjective "excellent" and so on, and stores the part of the title for use when generating the alternative title for the current article subsequently.

Optionally, if the target article only includes the first article, that is, the article with the quality degree greater than the first threshold value in the articles stored in the history library, the hot word may represent some commonly used high-quality words. Optionally, if the target article only includes the second article, that is, the article with the heat value greater than the second threshold value in the article published by the at least one social platform, the hot word may represent a word currently with a fire heat, and the word is easier to attract the attention of the user. Optionally, if the target article includes the first article and the second article, the sentence structure of the title may be obtained from the first article, and the hot word with a high frequency of occurrence may be obtained from the second article, so that not only the excellent titles of the historical articles may be drawn, but also the current hot spot may be caught, thereby attracting the attention of the user. It should be noted that the above sentence structure and the hot word obtaining manner are only examples, and the embodiment of the present invention does not limit this. For example, sentence structures of the titles of the first article and the second article and hot words in the first article and the second article can be obtained simultaneously.

And S104, generating alternative titles for the current article according to the keywords, the sentence pattern structure and the hot words.

In one embodiment, at least one alternative title is generated for the current article by adopting a deep learning algorithm according to the keywords of the current article, the acquired sentence pattern structure and the acquired hot words. Specifically, optionally, the keywords of the current article may include at least two words, and according to the semantics of the hot word and the semantics of the words included in the keywords, the hot word is used to replace the words of which the semantics are the same as those of the hot word in the at least two words, for example, if the current technology-based network hot word is AlphaGo, and the keywords of the current article include google, the google can be replaced by the parent company of AlphaGo, because the word of AlphaGo can attract the attention of the user more in the current period of time. Further, according to the part of speech of each participle in the sentence structure, assembling at least two terms after the replacement process into an alternative title, it should be noted that, in the process of assembling the alternative title, in order to make the sentence of the alternative title smooth, a part of terms may be modified appropriately, for example, if the keyword is "one day", then modified to "one day". Further optionally, if the keyword does not include the part of speech in the sentence structure, the high-frequency participle serving as the part of speech in the title may be searched from the stored title participles, and the high-frequency participle is used as the participle of the alternative title. For example, the current article is related to science and technology, the "introduction of the software development engineer doing AI by Google company" one day work life ", the analysis of the historical published articles results in a sentence structure of" adjective! The noun | turn | question? ", the hot word is AlphaGo; performing semantic analysis on the current article to obtain keywords: keywords such as daily | life | daily | google; after assembling the replaced words according to the sentence structure, the obtained alternative title may be "go to work every day at the AlphaGo parent", further combining the high frequency participles as the title and the "adjective! "and" question? ", the last generated alternative title may be" Bar extreme! Do you dislike doing so every day in the AlphaGo parent company? ".

It should be noted that the above-mentioned manner of generating the alternative titles is only an example, and other manners of generating the alternative titles may also be used, depending on the deep learning algorithm used.

Referring to fig. 2, a flow chart of another title generation method according to an embodiment of the present invention is shown. As shown in fig. 2, the method of the embodiment of the present invention may include the following steps S201 to S206.

S201, performing semantic analysis on a current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article;

step S201 in the embodiment of the present invention please refer to step S101 in the embodiment of fig. 1, which is not described herein again.

S202, acquiring a first article which belongs to the target category and has the quality degree greater than a first threshold value from articles stored in a historical library, wherein the quality degree is used for representing the quality degree of the target article;

in one embodiment, the articles stored in the historian are analyzed to determine a first article of the articles stored in the historian that belongs to the target category and has a goodness greater than a first threshold, the first article may include one or more articles, the article with a higher goodness indicating that the article is more popular with the user, and the article with a lower goodness indicating that the article is less popular with the user.

Optionally, at least one article belonging to the target category is first obtained from the articles stored in the history library, and if the target category is the science and technology category, at least one article of the science and technology category is screened from the articles stored in the history library. And then determining the quality of each article according to the reading quantity of each article and the interaction characteristics, wherein the interaction characteristics can include but are not limited to negative comment proportion of each article, collection quantity of each article, praise quantity of each article, forwarding share times of each article and the like.

As shown in fig. 4, that is, a schematic diagram of a high-quality historical headline model provided in the embodiment of the present invention is shown, as shown in the diagram, modeling training is performed on the reading number, the negative comment proportion, the collection number, the like number, and the forward sharing number of the article to obtain whether the article is high-quality, if the article is high-quality, the quality degree needs to be greater than a first threshold, and if the quality degree is high, the greater the quality degree is, the greater the reading number, the lesser the negative comment proportion, the greater the collection number, the greater the like number, and the greater the forward sharing number. And dividing at least one article screened out into a high-quality article and a low-quality article according to the quality degree, wherein the first article mentioned in the embodiment of the invention is the high-quality article in the figure 4.

S203, obtaining a second article which belongs to the target category and has a popularity value larger than a second threshold value in the articles published by at least one social platform, wherein the popularity value is used for representing the reading number of the target article.

In one embodiment, the articles published by at least one social platform in a recent period of time are analyzed to determine articles with a higher popularity value published by the at least one social platform, which may indicate that the articles are able to attract the attention of the user. Wherein, the at least one social platform may include, but is not limited to, microblogs, news sites, Baidu Fengyun charts, and the like.

Optionally, at least one article belonging to the target category in articles published by at least one social platform within a target duration range before the current time is obtained first, for example, a science and technology type article published in the last period is obtained. For the same event, different articles may be published on each social platform, so at least one article published on each social platform needs to be classified into at least one category, where the articles belonging to the same category have the same content, for example, if an article 1 is published for an event a on a microblog, an article 2 is published for an event a on a Baidu Fengyun chart, and an article 3 is published for an event a on Tencent news, the articles 1, 2, and 3 can be classified into one category. Then, for each category, the popularity values of the articles belonging to the category on the at least one social platform are accumulated to obtain the popularity value corresponding to the category. For example, if the popularity value of the article 1 for the event a on the microblog is a, the popularity value of the article 2 for the event a on the Baidu Fengyun chart is b, and the popularity value of the article 3 for the event a on the Tencent News is c, the popularity value a + b + c is used as the popularity value corresponding to the event a category. There is a heat value for each category, where the heat value may be the number of reads to the article. Finally, a target category with the heat value larger than the second threshold is screened from all categories, and all the articles belonging to the target category are the second articles with the heat value larger than the second threshold mentioned in the embodiment of the invention.

As shown in fig. 6, which is a schematic diagram of a hotspot mining model according to an embodiment of the present invention, data is crawled from each social platform by using a script automatic data crawling manner to monitor each large site, as shown in the figure, published articles are crawled from microblogs, large news sites, Baidu Fengyun charts, and the like, and the articles with higher heat values in a recent period of time are analyzed by performing classification processing according to publication time and heat values of each article and events targeted by each article.

S204, acquiring the title of the first article and acquiring a sentence structure of the title;

in an embodiment, after the first article with the quality degree greater than the first threshold is analyzed in step S202, the title of the first article is further obtained, the title is participled to obtain a sentence structure of the titles of the historically published high-quality articles, and the sentence structure of the titles of the high-quality articles is stored, as shown in the high-quality historical title model shown in fig. 4, the sentence structure, the punctuation mark and the participle of the obtained titles of the historically published high-quality articles are stored.

S205, acquiring the hot words of which the occurrence frequency is greater than a third threshold value in the second article.

In an embodiment, after the second article with the heat value greater than the second threshold is analyzed in step S203, a hot word with a relatively high occurrence frequency in the second article is further obtained, and the hot word can reflect an event with a relatively hot fire in a recent period of time and can attract the attention of the user most. And storing the acquired hot words.

And S206, generating alternative titles for the current article according to the keywords, the sentence pattern structure and the hot words.

In an embodiment, the step S201 is adopted to perform keyword processing on the content of the current article, the step S205 is adopted to obtain a sentence structure of the titles of the high-quality articles in the history library, the step S206 is adopted to obtain hot words in the articles published by each large social platform in the recent period, and further, at least one alternative title is generated for the current article according to the keyword, the sentence structure and the hot words, and a specific manner for generating the alternative title may refer to the description in the step S104, which is not described herein again.

As shown in fig. 3, which is a schematic diagram of a title generation model provided in an embodiment of the present invention, as shown in the figure, according to a title sentence structure of a historically published high-quality article trained by a high-quality historical title model, keywords of a current article extracted by an article content tag extraction model, and hot words in an article with a higher social platform popularity value in a recent period extracted by a hot spot mining model, an alternative title is generated by using a title assembly model and displayed to a user, and the user may select one title from at least one alternative title as a title of the current article.

In the embodiment of the invention, the target category of the current article and the keywords for representing the content of the current article are acquired, the first article which belongs to the target category and has the high quality degree higher than the first threshold value in the articles stored in the historical library is further acquired, the second article which belongs to the target category and has the heat value higher than the second threshold value in the articles published by the social platform is acquired, and the alternative titles attracting the attention of the user are generated for the current article by combining the sentence structure of the title of the first article and the hot words in the second article, so that the exposure rate and the reading amount of the current article are improved.

The title generating device provided by the embodiment of the invention will be described in detail below with reference to fig. 7 to 10. It should be noted that the apparatuses shown in fig. 7-10 are used for executing the method according to the embodiments of the present invention shown in fig. 1-6, and for convenience of description, only the parts related to the embodiments of the present invention are shown, and details of the technology are not disclosed, please refer to the embodiments of the present invention shown in fig. 1-6.

Fig. 7 is a schematic structural diagram of a title generation apparatus according to an embodiment of the present invention. As shown in fig. 7, the title generation device 1 according to the embodiment of the present invention may include: a semantic analysis unit 11, a first acquisition unit 12, a second acquisition unit 13, and a title generation unit 14;

a semantic analysis unit 11, configured to perform semantic analysis on a current article to obtain a target category to which the current article belongs and a keyword used for representing content of the current article;

The first acquiring unit 12 is configured to acquire a target article that belongs to the target category and has a popularity degree greater than a target threshold value from among historically published articles;

Alternatively, as shown in fig. 8, the first obtaining unit 12 may include a first article obtaining sub-unit 121 and a second article obtaining sub-unit 122;

a first article obtaining subunit 121, configured to obtain a first article that belongs to the target category and has a quality degree greater than a first threshold in the articles stored in the history library, where the quality degree is used to indicate a quality degree of the target article;

specifically, optionally, the first article obtaining subunit 121 is specifically configured to obtain at least one article belonging to the target category in the articles stored in the history library;

The second article obtaining subunit 122 is configured to obtain a second article that belongs to the target category and has a popularity value greater than a second threshold, where the popularity value is used to indicate a reading number of the target article, from among the articles published by the at least one social platform.

Specifically, optionally, the second article obtaining subunit 122 is specifically configured to obtain at least one article belonging to the target category from among articles published by the at least one social platform within a target duration range before the current time;

A second obtaining unit 13, configured to obtain a sentence structure of a title of the target article and a hot word in the target article;

Optionally, as shown in fig. 9, the second obtaining unit 13 may include a sentence structure obtaining subunit 131 and a hot word obtaining subunit 132;

a sentence structure obtaining subunit 131, configured to obtain a title of the first article, and obtain a sentence structure of the title;

and a hot word acquiring subunit 132, configured to acquire a hot word in the second article, where the occurrence frequency of the hot word is greater than a third threshold.

And a title generating unit 14, configured to generate an alternative title for the current article according to the keyword, the sentence structure, and the hot word.

Optionally, the sentence structure includes parts of speech of at least two participles of the title, and the keyword includes at least two words, as shown in fig. 10, the title generating unit 14 may include a replacing subunit 141 and a title assembling subunit 142;

a replacing subunit 141, configured to replace, by the hot word, a word with the same semantic as the hot word in the at least two words;

a title assembling subunit 142, configured to assemble the at least two replaced words into the alternative titles of the current article according to the parts of speech of each participle in the sentence structure.

For the concepts, explanations, and detailed descriptions related to the technical solutions provided in the embodiments of the present application and other steps related to the title generating device, reference is made to the descriptions of the foregoing methods or other embodiments, and no further description is given here.

An embodiment of the present invention further provides a computer storage medium, where the computer storage medium may store a plurality of instructions, where the instructions are suitable for being loaded by a processor and executing the method steps in the embodiments shown in fig. 1 to 6, and a specific execution process may refer to specific descriptions of the embodiments shown in fig. 1 to 6, which are not described herein again.

Fig. 11 is a schematic structural diagram of a server according to an embodiment of the present invention. As shown in fig. 11, the server 1000 may include: at least one processor 1001, such as a CPU, at least one network interface 1003, memory 1004, at least one communication bus 1002. Wherein a communication bus 1002 is used to enable connective communication between these components. The network interface 1003 may optionally include a standard wired interface, a wireless interface (e.g., a Wi-Fi interface). The memory 1004 may be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as at least one disk memory. The memory 1004 may optionally be at least one storage device located remotely from the processor 1001. As shown in fig. 11, the memory 1004, which is a kind of computer storage medium, may include therein an operating system, a network communication module.

In the server 1000 shown in fig. 11, a network interface 1003 is used for data communication with a user terminal; and the processor 1001 may be configured to call the program stored in the memory 1004 and specifically perform the following operations:

Optionally, the obtaining, by the processor 1001, a target article belonging to the target category and having a popularity degree greater than a target threshold from among the historically published articles includes:

Optionally, the obtaining, by the processor 1001, a first article which belongs to the target category and has a quality degree greater than a first threshold in the articles stored in the history library specifically includes:

Optionally, the obtaining, by the processor 1001, a second article that belongs to the target category and has a popularity value greater than a second threshold in the articles published by the at least one social platform specifically includes:

Optionally, the processor 1001 obtains a sentence structure of the title of the target article and a hot word in the target article, where the sentence structure includes:

Optionally, the sentence structure includes parts of speech of at least two participles of the title, the keyword includes at least two words, and the processor 1001 generates an alternative title for the current article according to the keyword, the sentence structure, and the hot word, specifically including:

It should be noted that, for a specific implementation process, reference may be made to specific descriptions of the method embodiments shown in fig. 1 to fig. 6, which are not described herein again.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and includes processes of the embodiments of the methods described above when the computer program is executed. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.

Claims

1. A title generation method, comprising:

performing semantic analysis on a current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article, wherein the keywords comprise at least two words;

obtaining a second article which belongs to the target category and has a popularity value larger than a second threshold value in articles published by at least one social platform, wherein the popularity value is used for representing the reading number of the target article;

acquiring a title of the first article, and acquiring a sentence pattern structure of the title, wherein the sentence pattern structure comprises the parts of speech of at least two participles of the title;

acquiring a hot word with the occurrence frequency larger than a third threshold value in the second article;

replacing words with the same semantic as that of the hot word in the at least two words by the hot word;

2. The method of claim 1, wherein obtaining a first article of the articles stored in the historian that belongs to the target category and has a goodness greater than a first threshold comprises:

3. The method of claim 1, wherein the obtaining a second article of the at least one social platform published article that belongs to the target category and has a popularity value greater than a second threshold comprises:

4. A title generation device, comprising:

the semantic analysis unit is used for carrying out semantic analysis on the current article to obtain a target category to which the current article belongs and keywords for representing the content of the current article, wherein the keywords comprise at least two words;

a title generating unit, configured to generate an alternative title for the current article according to the keyword, the sentence pattern structure, and the hot word;

the first acquisition unit includes:

the second article obtaining subunit is configured to obtain a second article that belongs to the target category and has a popularity value greater than a second threshold value among articles published by at least one social platform, where the popularity value is used to indicate a reading number of the target article;

the second acquisition unit includes:

a sentence pattern structure obtaining subunit, configured to obtain a title of the first article, and obtain a sentence pattern structure of the title, where the sentence pattern structure includes parts of speech of at least two participles of the title;

the hot vocabulary acquisition subunit is used for acquiring the hot vocabularies, the occurrence frequency of which is greater than a third threshold value, in the second article;

the title generation unit includes:

the replacing subunit is used for replacing the words with the same semantics as those of the hot vocabulary in the at least two words by adopting the hot vocabulary;

5. A computer storage medium having stored thereon a plurality of instructions adapted to be loaded by a processor and to perform the steps of:

6. A server, comprising: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the steps of: