CN113822077A

CN113822077A - Topic generation method and device

Info

Publication number: CN113822077A
Application number: CN202110786949.9A
Authority: CN
Inventors: 陈姿
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2021-07-12
Filing date: 2021-07-12
Publication date: 2021-12-21

Abstract

The embodiment of the application relates to the technical field of computers, and discloses a topic generation method and a topic generation device, wherein the method comprises the following steps: performing semantic recognition on one or more target comment contents of the multimedia data to obtain target semantic information; the topicality corresponding to the target comment content or the target semantic information meets a preset topicality condition, and the topicality is used for indicating the participation degree of the multimedia data; and generating the discussion topic of the multimedia data based on the target semantic information, thereby improving the generation efficiency of the topic.

Description

Topic generation method and device

Technical Field

The present application relates to the field of computer technologies, and in particular, to a topic generation method and apparatus.

Background

In order to increase the user engagement, most multimedia data platforms set a topic discussion zone, in which at least one topic can be included. However, the traditional topics are all edited manually, and a lot of human resources are consumed, so that the topic generation efficiency is low, and therefore how to improve the topic generation efficiency becomes a current research hotspot.

Disclosure of Invention

The embodiment of the application provides a topic generation method and device, which can improve the topic generation efficiency.

In one aspect, an embodiment of the present application provides a topic generation method, including:

performing semantic recognition on one or more target comment contents of the multimedia data to obtain target semantic information; wherein, the topicality corresponding to the target comment content or the target semantic information meets a preset topicality condition, and the topicality is used for indicating the participation degree of the multimedia data;

and generating the discussion topic of the multimedia data based on the target semantic information.

Correspondingly, the embodiment of the present application provides a topic generation apparatus, including:

the semantic recognition unit is used for performing semantic recognition on one or more pieces of target comment content of the multimedia data to obtain target semantic information; wherein, the topicality corresponding to the target comment content or the target semantic information meets a preset topicality condition, and the topicality is used for indicating the participation degree of the multimedia data;

and the generating unit is used for generating the discussion topic of the multimedia data based on the target semantic information.

Accordingly, the present application provides a topic generation apparatus comprising:

a processor adapted to implement one or more instructions;

a computer storage medium having stored thereon one or more instructions adapted to be loaded by the processor and to execute the above-described topic generation method.

Accordingly, an embodiment of the present application provides a computer storage medium, wherein the computer storage medium stores one or more instructions, and the one or more instructions are adapted to be loaded by the processor and execute the topic generation method.

Accordingly, embodiments of the present application provide a computer program product or a computer program, the computer program product comprising a computer program, the computer program being stored in a computer storage medium; the processor of the topic generation device reads the computer program from the computer storage medium, and the processor executes the computer program, so that the topic generation device executes the topic generation method described above.

According to the topic generation method provided by the embodiment of the application, the target semantic information is obtained by performing semantic identification on one or more target comment contents of the multimedia data, and then the discussion topic of the multimedia data is generated based on the target semantic information. In addition, as the exclusive discussion topic is generated for the discussion hotspot, the multimedia data can have more topic volume.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a schematic flowchart of a topic generation method provided in an embodiment of the present application;

fig. 2 is a schematic diagram of bullet screen content provided in an embodiment of the present application;

fig. 3 is a schematic flowchart of generating a discussion topic based on bullet screen content according to an embodiment of the present application;

fig. 4 is a schematic flow chart of a method for determining a target multimedia segment based on a discussion thread according to an embodiment of the present application;

fig. 5 is a schematic flow chart of still another method for determining a target multimedia segment based on a discussion thread provided in an embodiment of the present application;

fig. 6a is a schematic flowchart of face recognition according to an embodiment of the present application;

FIG. 6b is a schematic flow chart illustrating motion recognition according to an embodiment of the present disclosure;

fig. 7a is a schematic flowchart of determining a target multimedia segment matching a discussion thread according to an embodiment of the present application;

fig. 7b is a schematic flowchart of another process for determining a target multimedia segment matching a discussion thread provided in an embodiment of the present application;

fig. 7c is a schematic flowchart of another process for determining a target multimedia segment matching a discussion thread provided in an embodiment of the present application;

fig. 7d is a schematic flowchart of generating discussion topics based on the comment contents of the comment area according to the embodiment of the present application;

fig. 8 is a schematic diagram illustrating a method for triggering display of a target multimedia segment according to an embodiment of the present application;

fig. 9 is a schematic structural diagram of a topic generation device provided in an embodiment of the present application;

fig. 10 is a schematic structural diagram of a topic generation device provided in an embodiment of the present application.

Detailed Description

The topic generation method can automatically generate discussion topics for multimedia data, particularly generate discussion topics for multimedia data with high topicality so as to increase the topicality of the multimedia data, and the topicality is used for indicating participation degree of the multimedia data. The multimedia data with high topicality can refer to: multimedia data which arouses extensive attention and intense discussion of the masses on the media, wherein the multimedia data can be video data (such as videos of famous star participating programs), text data (such as character reports of extraordinary natural disasters) and audio data (such as broadcast propaganda of important activities); discussion topics may refer to: after brief and brief summarization is carried out on various comments published by the masses aiming at the multimedia data, the topic topics (or called topic points) are obtained. For example, suppose that a video a is a video of a dance program performed by children on site when a juxtapose of international giant star, a, participates in a public welfare activity of a building school, and a large number of net friends post comments in a comment area of the video a after watching the video a; in this case, if a large number of comments in the comment area are all jumping well on dancing in favor of the international Jupiter Xiao a, the discussion topics for this multimedia data (i.e.: video A) in the comment area may be: dancing of little a is too long; if a large number of comments in the comment area are called to learn from Xiao a by more people to participate in public welfare activities, the discussion topic of the multimedia data for the comment area may be: and the small a are participated in the public welfare activities together.

The following describes a general principle of a topic generation method provided in an embodiment of the present application, where the topic generation method mainly obtains target semantic information by performing semantic recognition on one or more pieces of target comment content of multimedia data, and further generates a discussion topic based on the target semantic information. In this case, the target semantic information may be understood as: the semantics expressed by most of the comment contents in the comment contents can further specify that the topicality of the target semantic information meets a preset topicality condition; optionally, the one or more mentioned target comment contents may also specifically refer to a specific one or more comment contents, and the topicality of each comment content in the specific one or more comment contents satisfies a preset topicality condition, for example, comment contents with a large number of prawns or a large browsing amount may be understood as a comment content with a high participation degree in the multimedia data, that is, a comment content whose topicality satisfies a preset topicality condition. It should be noted that the above mentioned preset topic condition may include, but is not limited to, any one of the following: the topicality is arranged from high to low and then positioned at the Nth position (such as the 1 st position, the 2 nd position and the 3 rd position), the topicality meets the topicality threshold value, the topicality is arranged from high to low and then positioned at the front N position (such as the front 2 th position and the front 3 rd position), and the like, wherein N is a positive integer and is less than or equal to the number of one or more comments.

Based on the above description, it can be understood that, because the topic generation method provided by the embodiment of the present application generates the discussion topic based on the semantic information of the target comment content after performing semantic identification on the target comment content of the multimedia data, the embodiment of the present application can avoid manual participation in generation of the discussion topic (for example, avoid manual editing of the discussion topic), realize automatic generation of the discussion topic, and save human resources to a certain extent.

In one embodiment, the topic generation method may be performed by a topic generation device, and the topic generation device may be a server. Specifically, the server may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a Network service, cloud communication, a middleware service, a domain name service, a security service, a Content Delivery Network (CDN), a big data and artificial intelligence platform, which is not limited in this embodiment of the present application.

Referring to fig. 1, fig. 1 is a schematic flowchart of a topic generation method provided by an embodiment of the present application, where the topic generation method may be executed by the above-mentioned topic generation apparatus, as shown in fig. 1, the method includes steps S101 to S102:

s101, performing semantic recognition on one or more target comment contents of the multimedia data to obtain target semantic information.

The target comment content can be bullet screen content corresponding to the multimedia data, and the bullet screen content refers to: the method comprises the following steps of watching the popped commenting subtitles when the multimedia data is watched on line through the internet, wherein the barrage content can be displayed on a playing interface of the multimedia data in an overlapping manner, the time length of the barrage content displayed on the playing interface of the multimedia data in the overlapping manner is preset time length, and the preset time length is generally not more than the total playing time length of the multimedia data, such as: when the total playing time of the multimedia data is 20 seconds, the preset time may be as follows: 3 seconds, 5 seconds, etc. As shown in fig. 2, the bullet screen content 201 stops being displayed on the playing interface of the multimedia data after a preset time period. It should be noted that, with the playing of the multimedia data, the bullet screen content displayed on the terminal device may be changed, where the change includes, but is not limited to, any one or more of the following: as shown in fig. 2, the bullet screen content 201 and the bullet screen content 202 are both displayed in a superimposed manner on the playing interface of the multimedia data, after a period of time (e.g., a preset time duration) elapses, the bullet screen content 202 stops being displayed in a superimposed manner on the playing interface of the multimedia data, and the number of the bullet screen content displayed on the terminal device changes, where the terminal device is a device for playing the multimedia data, that is, a device for displaying the playing interface of the multimedia data. Optionally, the target comment content may also be comment content in a comment area of the multimedia data, where the comment content in the comment area may include, but is not limited to, any one or more of the following: the system comprises picture comment content, video comment content, text comment content and audio comment content.

In specific application, the topic generation device can adopt a natural language identification model to perform semantic identification on the comment content, and it needs to be explained that the natural language identification model is constructed based on a natural language processing technology; natural Language Processing (NLP) is a technology for studying Natural Language of human being (i.e. Language used by people in daily life) processed by computer, and is also an important direction in the fields of computer science and artificial intelligence. In addition, natural language processing is a science integrating linguistics, computer science and mathematics, and natural language processing technologies generally include technologies such as text processing, semantic understanding (or semantic recognition), machine translation, robot question answering and knowledge mapping. Based thereon, it will be appreciated that the natural language processing may include natural language recognition, which may include: the method comprises the steps of syntactic semantic analysis, information extraction, text mining, machine translation, information retrieval and the like, so that the semantic identification of the target comment content can be realized by adopting a natural language identification model, and the semantic information of the target comment content can be obtained.

In an embodiment, as can be known from the foregoing, the one or more pieces of target comment content may refer to a plurality of pieces of comment content, in which case, the one or more pieces of comment content may be all pieces of comment content of the multimedia data (e.g., all pieces of comment content in a comment area of the multimedia data, or all bullet screen content of the multimedia data), or may be a plurality of pieces of comment content obtained after removing irrelevant comment content from all pieces of comment content of the multimedia data, for example, assuming that there are 50 pieces of comment content in a comment area of the gourmet video a, where 1 piece of comment content a is a piece of comment content for a certain cosmetic, the comment content a may be considered as a piece of comment content unrelated to the gourmet video a, and thus the corresponding one or more pieces of target comment content of the gourmet video a are 49 pieces of comment content after removing the comment content a. Based on this, the topic generation device performs semantic recognition on one or more pieces of target comment content of the multimedia data, and when obtaining target semantic information, the topic generation device may specifically include the following steps: the topic generation equipment firstly obtains one or more pieces of target comment content of the multimedia data for semantic recognition to obtain one or more pieces of semantic information, and after the topic generation equipment obtains the one or more pieces of semantic information, the number of comment content corresponding to similar semantic information in the one or more pieces of semantic information can be obtained; further, if the number of the comment contents corresponding to the target similar semantic information meets a first preset number condition, it is determined that the topicality corresponding to the target similar semantic information meets a preset topicality condition, and then the topicality generating device can take the target similar semantic information as the target semantic information. Wherein, similar semantic information can be understood as: one or more similar semantic information, and when the similar semantic information comprises a plurality of semantic information, the similarity between any two semantic information is greater than a similarity threshold. The target similar semantic information can be understood as: and the number of the corresponding comment contents meets the similar semantic information of a first preset number condition. The first predetermined number of conditions may include, but is not limited to, any of the following: the number is the largest, the number meets a preset number threshold, the number is arranged at the front P positions according to the number after at least increasing the number, and the like, wherein P is a positive integer, and is larger than zero and smaller than or equal to the number of the similar semantic information.

The topic generation device obtains the number of comment contents corresponding to similar semantic information in one or more pieces of semantic information, and the method can be specifically realized by the following method: the topic generation equipment can determine one or more semantic information groups from one or more pieces of semantic information, the similarity between the semantic information contained in each semantic information group is greater than a preset proportion threshold, and each semantic information group can indicate one piece of similar semantic information; further, the topic generation device may obtain the sum of the number of the comment contents corresponding to all the semantic information in each semantic information group, and use the sum of the number of the comment contents corresponding to the semantic information group as the number of the comment contents corresponding to the similar semantic information indicated by the semantic information group. Since one or more semantic information sets can be determined, and each semantic information set indicates one similar semantic information, it is easy to understand that the number of similar semantic information sets is the number (i.e., one or more) of semantic information sets. Then based on this, the target similar semantic information can be understood as: and the sum of the quantities meets similar semantic information indicated by the semantic information group of the first preset quantity condition.

Optionally, the topic generation device may perform semantic identification on one or more pieces of target comment content of the multimedia data to obtain one or more pieces of semantic information, and after the topic generation device obtains the one or more pieces of semantic information, the number of comment content corresponding to the same semantic information in the one or more pieces of semantic information may be obtained; further, if the number of the comment contents corresponding to the semantic information meets a first preset number condition, it is determined that the topicality corresponding to the semantic information meets a preset topicality condition, and then the topic generation device can use the semantic information as the target semantic information. The first predetermined number of conditions may include, but is not limited to, any of the following: the number is the largest, the number meets a preset number threshold value, the number is arranged at the top M bits according to the number after at least more sorting, M is a positive integer, and M is larger than zero and smaller than or equal to the number of the semantic information.

In order to facilitate clear understanding of the topic generation method provided by the present application, taking the bullet screen content as the target comment content as an example, a specific method for obtaining the target semantic information when one or more pieces of target comment content are multiple pieces of comment content by the topic generation device is described in detail below: assuming that the first preset quantity condition is the maximum quantity, if the number of the bullet screen contents corresponding to the multimedia data A is 100, after the topic generation equipment carries out semantic understanding on each bullet screen content, 50 bullet screen contents with semantic information of XX dancing good commander are obtained, 30 bullet screen contents with semantic information of XX dancing good standard are obtained, and 20 bullet screen contents with semantic information of XX stature ratio good; then, since both the "XX dancing handsome" and the "XX dancing action good standard" have the two keywords "XX" and "dance", it may be exemplarily considered that the similarity between the semantic information "XX dancing handsome" and the semantic information "XX dancing action good standard" is greater than the similarity threshold, and then the topic generation device may add the two semantic information "XX dancing handsome" and "XX dancing action good standard" to the same semantic information group (assumed to be semantic information group a), and add the semantic information "XX stature ratio is too good" to the other semantic information group (assumed to be semantic information group b). Further, the topic generation device may obtain the number of the barrage contents corresponding to each semantic information group, and it is easy to see that, the total number of the barrage contents corresponding to the semantic information included in the semantic information group a is 80(50+30 is 80), and the total number of the barrage contents corresponding to the semantic information included in the semantic information group b is 20, and further, it may be determined that the total number 80 of the comment contents corresponding to the semantic information group a satisfies a first preset number condition, so that the topic generation device may use the semantic information included in the semantic information group a as the target semantic information, that is, the target semantic information is: "XX dances well in handsome" and "XX dances well in standard". It should be noted that, similarly, when one or more pieces of target comment content are pieces of comment content in the comment area, the topic generation device may also obtain the target semantic information in the manner in the above example, which is not described herein again.

In yet another embodiment, as can be seen from the foregoing, the one or more pieces of target comment content may also refer to a piece of comment content, in which case, the one or more pieces of target comment content may be a piece of comment content of all the comment contents of the multimedia data whose topicality satisfies a preset topicality. Any comment content topicality in all comment contents of the multimedia data can be obtained based on the interactive information of any comment content, and the interactive information can include, but is not limited to, any one of the following: the number of praise, the number of comments and the number of browsing times. Illustratively, the topicality can be specified to have a positive correlation trend with the size of the interactive information, such as: the higher the browsing frequency is, the higher the corresponding topicality is; optionally, it may also be specified that the topicality is in a positive correlation with the growth speed of the interactive information within a preset time period, such as: within the preset time length, the topic degree corresponding to the comment content is higher as the increase of the praise number is more. Based on this, the topic generation device performs semantic recognition on one or more pieces of target comment content of the multimedia data, and when obtaining target semantic information, the topic generation device may specifically include the following steps: the topic generation equipment firstly acquires the interactive information of each piece of comment content in one or more pieces of comment content of the multimedia data, and then determines the corresponding topic degree of each piece of comment content according to the interactive information of each piece of comment content; further, the topic generation device may determine, in one or more pieces of comment content, target comment content whose topicality satisfies a preset topic condition, perform semantic identification on the target comment content to obtain semantic information of the target comment content, and then the topic generation device may use the semantic information of the target comment content as the target semantic information. The target comment content with the topicality meeting the preset topicality can refer to: the comment content with the largest number of praise, comment number or browsing times, or the comment content with the largest numerical value obtained by performing arithmetic operation (for example, weighted summation or mean value) on at least two of the praise, comment number and browsing times, or the comment content with the largest numerical value obtained by performing arithmetic operation on at least two of the praise, comment number and browsing times, wherein L is a positive integer, and is greater than zero and less than the number of one or more comment contents of the multimedia data.

Similarly, in order to facilitate clear understanding of the topic generation method provided by the present application, taking an example that the topicality of the comment content and the praise number of the comment content are in a positive correlation trend, a specific method for the topic generation device to obtain the target semantic information when one or more target comment contents are a piece of comment content will be described in detail: assuming that the preset topic condition is the highest topic degree, if 3000 comment contents correspond to the multimedia data, the topic generation device may use the comment content with the highest praise number (assumed as comment content a) among the 3000 comment contents as the comment content whose topic degree meets the preset topic condition, so as to obtain the target comment content; further, the topic device may perform semantic recognition on the comment content a to obtain semantic information of the comment content a, and then the topic device may use the semantic information of the comment content a as target semantic information.

It should be noted that, in practical application, the target comment content may be any one of video comment content, audio comment content, picture comment content, and text comment content, which is not limited in this application. When the target comment content is video comment content, audio comment content, or picture comment content, the topic generation device may acquire semantic information of the target comment content in the following manner: the topic generation device firstly identifies the content of the target comment content to obtain text information corresponding to the target comment content, and then carries out semantic identification on the text information to obtain semantic information of the target comment content. Wherein, the content identification can be understood as: recording the information expressed by the target comment content by using the character content, and then, the text information may refer to: and the text content can be used for describing the semantic meaning expressed by the target comment content. For example, the content of the video comment content may be identified in a video understanding manner, the content of the audio comment content may be identified in a voice identification manner, and the content of the picture comment content may be identified in an image identification manner, which is not limited in the present application. For example, assume that the target comment content is video comment content, and the video comment content includes a video of dancer a; if the video characteristics obtained after the topic generation device carries out video understanding on the video are as follows: dancer a dances and the audience is applause. Then the text information corresponding to the video comment content may be: text content describing the scene of dancer a, such as: "dancer a learns by self and dances in four seats flawlessly". Then, as can be understood based on the above description, when the target comment content is the text comment content, the topic generation device may perform semantic recognition on the text comment content directly using a natural language recognition model to obtain semantic information of the text comment content.

And S102, generating the discussion topic of the multimedia data based on the target semantic information.

The target comment content corresponding to the target semantic information may be part or all of one or more pieces of target comment content. Based on this, the topic generation device may generate the discussion topic by: the topic generation equipment carries out word segmentation processing on all target comment contents corresponding to the target semantic information to obtain a plurality of keywords; at least one keyword can be obtained after word segmentation processing is carried out on each piece of comment content, and the same keyword can be obtained after word segmentation processing is carried out on different comment contents; further, the topic generation device can perform semantic understanding on each keyword in the plurality of keywords to obtain keywords with similar semantic information, and determine a target keyword from the keywords with similar semantic information; further, the topic generation device may generate a discussion topic based on all the determined target keywords, the discussion topic including at least all the determined target keywords, and the discussion topic may include, for example: all the determined target keywords and other words (such as adjectives, adverbs, prepositions and the like) for modifying all the determined target keywords.

In one embodiment, the target keyword may be the most numerous same keywords among similar keywords, and then the discussion topic at least includes all the obtained most numerous keywords mentioned above. In this case, the step of the topic generation device generating the discussion topic is explained in detail below with reference to a specific example: assuming that all target comment contents corresponding to the target semantic information are: comment content a "the small a sings are also too good at the bar", comment content B "i like too good at the small a singing, super-cured", and comment content C "what the fairy voice is, the singing sound is so cured". Then, the topic generation device may perform word segmentation on the comment content a, the comment content B, and the comment content C to obtain one or more keywords, and if the topic generation device performs word segmentation on the comment content a that "a sing is too good for listening" the obtained keywords are "a", "sing", "good for listening"; after the topic generation device carries out word segmentation processing on the comment content B 'I likes to hear little a singing and super cure too much', the obtained keywords are 'little a', 'singing' and 'cure'; after the topic generation device carries out word segmentation processing on the comment content C, namely the comment content C, which is the sound of the immortal throat and the singing sound is cured, the obtained keywords are as follows: "the sound of the throat of the immortal", "the sound of singing", "cure"; then, if the topic generation device performs word segmentation processing on all target comment contents corresponding to the target semantic information, one or more groups of keywords similar to the semantic information are obtained as follows: a group a { "small a", "small a" }, a group b { "singing", "singing sound" }, a group c { "hearing", "curing", "healing", "immortal throat sound" }; then, the topic generation device may obtain the number of the same keywords in each group of similar keywords, to obtain: the number of "small a" in group a is 2; the number of singing in the group b is 2, and the number of singing sounds is 1; in group c, the number of "good hearing" is 1, the number of "healing" is 2, and the number of "nervous voice" is 1; then, it can be seen that the most numerous keywords of all similar keywords include: "Xiao a", "sing", "cure". Thus, the topic generation device may generate the discussion topic based on "small a", "singing", "curing", and exemplarily, the discussion topic may include only these 3 keywords, such as: curing the small singing; these 3 keywords and other words (adjectives, verbs, prepositions, etc.) describing the keywords may also be included, such as: people praise little singing a to cure the heart of people, and the application does not limit the heart.

In another embodiment, the target keyword may be a keyword with the lowest word sense among keywords with similar semantic information, and the lowest word sense may be understood as: the word meaning covers a smaller range, and the referred things are more specific; for example: the semantic information of the three keywords including dance, street dance and thunderbolt dance is 'dance', but the thunderbolt dance belongs to one of the street dance, and the street dance also belongs to one of the dance, so that the thunderbolt dance is the word with the lowest word meaning in the keywords with similar semantic information, and then, when the topic generation device is specifically applied, the thunderbolt dance can be used as a target keyword.

In an embodiment, based on the above description related to steps S101 to S102, referring to fig. 3 as an example, a specific manner of generating a discussion topic of multimedia data by a topic generation device based on barrage content of the multimedia data, as shown in fig. 3, the topic generation device may first obtain all barrage content corresponding to the multimedia data, then understand the obtained barrage content (i.e. semantic recognition), and perform content aggregation on the barrage content with similar semantic information, where the content aggregation may be understood as: dividing bullet screen contents with similar semantic information together; then, based on this, the topic generation device can determine the weight of the semantic information based on the aggregated content (i.e. the number of bullet screen contents of similar semantic information), then, further, the topic generation device can prioritize the semantic information according to the weight of the semantic information, exemplarily, the more the number of bullet screen contents of similar semantic information is, the more the weight of the similar semantic information is, wherein the weights of the semantic information contained in the similar semantic information are the same, then based on this, the topic generation device can preferentially adopt the bullet screen contents with the maximum weight of the semantic information to generate discussion topics; such as: if 5 bullet screen contents of the similar semantic information a, 6 bullet screen contents of the similar semantic information b and 2 bullet screen contents of the similar semantic information c are provided, the similar semantic information is arranged according to the sequence of the weights from big to small: similar semantic information b > similar semantic information a > similar semantic information c, the topic generation device may generate the discussion topic based on the similar semantic information b.

According to the topic generation method provided by the embodiment of the application, the target comment content of the multimedia data is subjected to semantic recognition, then the similar semantic information with the number of the target comment content of the similar semantic information meeting a first preset number threshold value is used as the target semantic information, and the discussion topic of the multimedia data is generated based on the target semantic information. Therefore, the topic generation method provided by the embodiment of the application is generated based on the topic points with higher discussion volume, and compared with the traditional way of manually editing the discussion topics, the topic generation method provided by the embodiment of the application more accurately positions the discussion hotspots related to the multimedia data, the generated discussion topics have more objectivity, and the discussion topics can be ensured to be continuously concerned, so that more topic volumes are brought to the multimedia data corresponding to the discussion topics. Further, it can be understood that the embodiment of the application avoids artificial editing of discussion topics, so that the efficiency of topic generation is improved to a certain extent, and human resources are saved.

Based on the related description of the above embodiment, the topic generation method provided by the embodiment of the application may further associate and store the generated discussion topic and the target multimedia segment matched with the discussion topic after the topic generation device generates the discussion topic, so that the user may directly view the target multimedia segment related to the discussion topic on the topic discussion page, and thus the user may more quickly know the discussion content of the discussion topic, and the user experience when the user participates in the topic discussion is improved to a certain extent. Wherein, the target multimedia segment can be understood as: in the multimedia data, the multimedia segment strongly related to the discussion topic, so-called strongly related to the discussion topic can be understood as follows: the characteristics (such as pictures, sounds, characters and the like) indicated by the keywords corresponding to the discussion topics exist; for example, assuming the discussion topic is "Xiao a dancing and looking", then the multimedia segments that are strongly related to "Xiao a dancing and looking" can be: the multimedia data includes multimedia fragments of pictures dancing with a. In addition, it should be noted that, during the process of viewing the target multimedia segment, the user may also selectively view other multimedia segments or the entire multimedia data in the multimedia data in which the target multimedia segment is located. In practical applications, the association of the discussion topic with the target multimedia segment may be implemented by the above-mentioned topic generation device, and may also be implemented by other devices (e.g. association devices), where the above-mentioned association devices may be servers, server clusters, etc., and the present application embodiment does not limit this.

Referring to fig. 4, fig. 4 is a schematic flowchart of a method for determining a target multimedia segment based on a discussion thread according to an embodiment of the present application, which may be applied to a scene in which one or more pieces of target comment content are bullet screen content; the method shown in fig. 4 is explained in detail next, taking as an example that the topic generation device realizes the association of the discussion topic with the target multimedia segment; as shown in fig. 4, when associating the discussion topic with the target multimedia segment, the topic generation device specifically includes steps S401 to S406:

s401, obtaining the playing time length of the multimedia data.

The multimedia data may include, but is not limited to, any of the following: video data, audio data, text data, picture data. When the multimedia data is video data or audio data, the playing duration may refer to: the duration of the video data or audio data being played on the device; when the multimedia data is text data or picture data, the playing duration can be understood as: a data size of the text data or the picture data, or a total display time length of the text data or the video data. For convenience of description, the topic generation method provided by the present application is explained in detail by taking multimedia data as video data as an example in the present application without special description.

S402, dividing the playing time into a plurality of playing time periods.

In specific application, the topic generation device can uniformly divide the playing time into a plurality of playing time periods to obtain a plurality of playing time periods with consistent time; optionally, the topic generation device may also divide the playing time length non-uniformly into a plurality of playing time periods according to the actual demand, to obtain a plurality of playing time periods with inconsistent time lengths, where it should be noted that the plurality of playing time periods with inconsistent time lengths may refer to: in the multiple playing time periods, the durations of any two playing time periods are not consistent; it may also mean: in the multiple playing time periods, there is inconsistency of durations of any two playing time periods, which is not limited in the present application. Without special explanation, the present application will be described in detail by taking the case that the durations of the playing time periods in the plurality of playing time periods are consistent.

S403, acquiring the number of the segment comment contents of the multimedia segment played in each playing time period in the plurality of playing time periods.

In a specific embodiment, the topic generation device may first obtain time information of each piece of target comment content in one or more pieces of target comment content in a multimedia data playing process; and then determining the segment comment contents of the multimedia segments played in each playing time period according to the time information of each piece of target comment content and each playing time period, and further, the topic generation equipment can acquire the number of the segment comment contents of the multimedia segments played in each playing time period. The multimedia clips played in each playing time period are the multimedia clips of the multimedia data; the segment comment content is: the semantic information is comment content of the target semantic information, so that the comment content of the segment can be understood to be part or all of one or more pieces of target comment content corresponding to the multimedia data; the time information may include, but is not limited to, any of the following: it should be noted that, the display time and the release time mentioned herein are both for the playing time duration of the multimedia data, that is: the display time can be the time corresponding to the playing time when the bullet screen content is displayed in the playing process of the multimedia data; the release time may be a specific play time point in the play duration of the multimedia data.

The following describes a specific implementation manner of step S403 in detail with reference to a specific example. Assuming that the playing time of the multimedia data a is 20 minutes and the time of each playing time period is 5 minutes, after the topic generation device divides the playing time of the multimedia data by time periods, 4 playing time periods can be obtained, which are respectively: 0-5 minutes, 6-10 minutes, 11-15 minutes, 16-20 minutes; based on this, if the time information is the release time, when the multimedia data a corresponds to the bullet screen content with 10 pieces of semantic information as the target semantic information, where the release time of 3 pieces of bullet screen content is that the multimedia data a is played to 4 minutes and 30 seconds, the release time of 6 pieces of bullet screen content is that the multimedia data is played to 6 minutes and 01 seconds, and the release time of 1 piece of bullet screen content is 1 minute and 25 seconds, it is easy to understand that the number of the segment comment content in the playing time period of 0-5 minutes is 4, the number of the segment comment content in the playing time period of 6-10 minutes is 6, and the number of the segment comment content in the playing time period of 11-15 minutes and the playing time period of 16-20 minutes are both 0. Correspondingly, if the time information is the display time, when the multimedia data a corresponds to 10 bullet screen contents with semantic information as the target semantic information, where the display time of 3 bullet screen contents is 4 minutes 50 seconds to 5 minutes 20 seconds, the release time of 6 bullet screen contents is from multimedia data playing to 6 minutes 01 seconds to 6 minutes 31 seconds, and the release time of 1 bullet screen content is from 9 minutes 25 seconds to 10 minutes 01 seconds, it is not difficult to understand that the number of the segment comment contents in the playing time period of 0-5 minutes is 3, the number of the segment comment contents in the playing time period of 6-10 minutes is 10 (3+6+1 is 10), the number of the segment comment contents in the playing time period of 11-15 minutes is 1, and the number of the segment comment contents in the playing time period of 16-20 minutes is 0.

S404, determining the number of the comment contents of the segments meeting the second preset number condition.

Wherein the second preset number condition may include any one of: the number is the largest, the number exceeds a preset number threshold, and the number is arranged from big to small and then positioned at the front Q position (Q is a positive integer, and Q is larger than zero and is smaller than or equal to the number of the playing time period). After the topic generation device determines the number of the segment comment contents meeting the second preset number condition, the playing time period of the segment comment contents can be used as the time period when the bullet screen contents appear most densely.

S405, according to the time information of the segment comment content with the earliest release time in the determined segment comment contents, determining the initial position of the target multimedia segment matched with the discussion topic in the multimedia data.

Wherein the starting position can be understood as: in the playing duration of the multimedia data, the specific playing start time of the target multimedia segment is as follows: if the target multimedia segment starts playing in the 2 nd, 30 th second of the playing duration, the starting position of the target multimedia segment is considered as follows: 2 nd minute and 30 seconds. In practical application, since the actual appearance time of the bullet screen content is not necessarily the time of the beginning of the playing time period, in order to obtain a more accurate starting position of the target multimedia segment, the topic generation device may obtain the appearance time of a first segment comment content in the playing time period, where the first segment comment content refers to: and in all the segment comment contents corresponding to the target multimedia segment, the segment comment content with the earliest release time.

S406, establishing an association relationship between the discussion topic and the starting position, and storing the discussion topic and the starting position associated with the discussion topic.

The discussion topic and the starting position associated with the discussion topic can be stored in a blockchain, the blockchain is composed of one or more servers corresponding to the terminal device of the user for posting the comment content and the topic generation device, and each of the terminal device of the user and the one or more servers is a node on the blockchain. It can be understood that by storing the discussion thread in association with the start position of the target multimedia segment, the topic generation device can trigger to play the target multimedia segment or trigger to display a play entry of the target multimedia segment when detecting the browsing of the content related to the discussion thread by the user.

The topic generation method provided by the embodiment of the application determines the target multimedia segment matched with the discussion topic in the multimedia data, and stores the discussion topic in association with the starting position of the target multimedia segment, so that a user can quickly jump to the multimedia segment corresponding to the starting position on the discussion page where the discussion topic is located for watching, and the user experience can be improved to a certain extent.

Referring to fig. 5, fig. 5 is a schematic flowchart of still another method for determining a target multimedia segment based on a discussion thread provided in an embodiment of the present application, which may be applied to a scenario in which one or more target comment contents are comment contents in a comment area, or one or more target comment contents are bullet screen contents; as shown in fig. 5, in this case, the topic generation device, when associating the discussion topic with the target multimedia segment, specifically includes steps S501 to S503:

s501, determining a target multimedia segment matched with the discussion topic in the multimedia data.

In one embodiment, it is understood that the time length of the multimedia segment in which the target comment content appears most densely may be longer than the time length corresponding to each playing time period, and therefore, in order to obtain a complete multimedia segment that is strongly related to the discussion topic, when the target comment content is the barrage content, the manner in which the topic generation device determines the target multimedia segment may include the following steps: (1) acquiring the playing time length of multimedia data; (2) dividing the playing time into a plurality of playing time periods; (3) acquiring the number of the segment comment contents of the multimedia segments played in each playing time period in a plurality of playing time periods; the multimedia clips played in each playing time period are multimedia clips of multimedia data, and the semantic information of the clip comment content is target semantic information; (4) determining first multimedia fragments corresponding to the fragment comment contents of which the number meets a second preset number condition; the first multimedia segment may be: and the number of the multimedia fragments with the same bullet screen content as the target semantic information meets a second preset number threshold. (5) And determining a second multimedia segment adjacent to the first multimedia segment in the multimedia data, wherein the second multimedia segment is matched with the discussion topic corresponding to the first multimedia segment. (6) And obtaining the target multimedia segment based on the first multimedia segment and the second multimedia segment.

It should be noted that, for the related embodiments of step (1) to step (4), reference may be made to the related descriptions of step S401 to step S404 in fig. 4, which is not described herein again. Furthermore, the above mentioned second multimedia segment matching the discussion thread can be understood as: the second multimedia segment comprises the characteristics (such as human face characteristics, action characteristics and the like) of each keyword corresponding to the discussion topic; then, further, the second multimedia segment can be understood as: and other multimedia fragments including the characteristics of each keyword corresponding to the discussion topic except the first multimedia fragment in the multimedia data. The playing time of the second multimedia clip may be earlier than the playing time of the first multimedia clip, that is, the second multimedia clip is: the multimedia clip corresponding to the playing time period before the playing time period of the first multimedia clip; or the playing time of the second multimedia clip may be later than the playing time of the first multimedia clip, that is, the second multimedia clip is: the multimedia clip corresponding to the playing time period after the playing time period of the first multimedia clip; or the playing time of the first multimedia sub-segment in the second multimedia segment may be earlier than the playing time of the first multimedia segment, and the playing time of the second multimedia sub-segment in the second multimedia segment is later than the playing time of the first multimedia segment, that is: the second multimedia clip includes both a multimedia clip whose play period is before the play period of the first multimedia clip and a multimedia clip whose play period is after the play period of the first multimedia clip. This is not limited by the present application.

In yet another embodiment, when the target comment content is the comment content of the comment area, the topic generation device may determine the target multimedia segment in the following manner: firstly, the topic generation equipment carries out word segmentation processing on a discussion topic to obtain one or more keywords; a target multimedia segment is then determined in the multimedia data that matches the one or more keywords. In practical application, when the topic generation device determines a target multimedia segment matching one or more keywords in the multimedia data, the topic generation device may specifically include the following steps: acquiring an identification material of each keyword in one or more keywords, and acquiring a material characteristic of each identification material, wherein the identification material is used for indicating the characteristic corresponding to the keyword, and the identification material is used for indicating the characteristic corresponding to the keyword; the topic generation equipment can identify each identification material to obtain the material characteristics of the identification material; further, after the topic generation device obtains the material features of each identification material, feature recognition may be performed on the multimedia data based on the material features of each identification material, and a third multimedia segment is determined in the multimedia data, where the third multimedia segment includes the material features of the identification material of each keyword in the one or more keywords, and after the third multimedia segment is determined, the topic generation device may use the third multimedia segment as a target multimedia segment matched with the one or more keywords.

The identification material may include, but is not limited to: human face materials (e.g. XX human face images), action materials (e.g. videos including the action of a thunderbolt dance, motion pictures including a waving action, etc.), audio materials (e.g. accompaniment of XX song, XX animal sound, etc.), text materials (lyric of XX song, title of XX person, etc.), static real object materials (e.g. XX flower images, mountain images, field images, train station images, etc.), etc. The topic generation equipment can adopt different identification methods to identify the characteristics of different identification materials, and then explains the identification modes of several common identification materials: (1) if the identification material is a face material (for example, a face image of a main person mentioned in the multimedia data), the topic generation device may perform face identification on the video data (one type of multimedia data) according to the face material to obtain a face feature of the face material, and then determine a video segment of the person indicated by the face feature according to the face feature. For an exemplary specific process of performing face recognition by the topic generation device, as shown in fig. 6a, the topic generation device first pre-processes a face material and each face image summarized based on video data, then performs feature extraction on the face material and each face image in the video data, and then compares features of the face material and features of the face image to determine a face image matched with the features of the face material. Illustratively, the topic generation device can adopt a deep learning-based feature extraction method (such as a Convolutional Neural Network (CNN) algorithm) to extract features of the face materials. The CNN is an artificial neural network based on deep learning theory, which mainly utilizes weight sharing to reduce the parameter expansion problem in a common neural network, uses convolution kernel to perform convolution operation on input data in the forward calculation process, and uses the obtained result as the output of the layer through a nonlinear function, wherein the layer is called a convolution layer, a downsampling layer can be generated between the convolution layer and the convolution layer, the downsampling layer is mainly used for acquiring the invariance of local features, and the reduction of the scale of a feature space is generally followed by the convolution layer and the downsampling layer and is a fully-connected neural network for final identification. (2) If the identification material is an action material (such as a motion video of thunderbolt dancing, a motion video of waving hands and the like), the topic generation equipment can firstly identify the action of the video data according to the action material to obtain the action characteristics of the action material, and then determine a video segment of the video data, which comprises the action indicated by the action characteristics, according to the action characteristics. For an exemplary process of extracting features of the action material by the topic generation device, see fig. 6b, the topic generation device first extracts features of the action material, then extracts features of the video data to obtain feature representations of the action material, and then can identify and understand actions of the video data based on the feature representations. Optionally, the topic generation device may further store the action features corresponding to the action material in the video data and the determined video segments in an associated manner (for example, store the action features and the determined video segments in a database of the server), so that the topic generation device may directly obtain the video segments corresponding to the action from the database when the action corresponding to the action material needs to be identified next time, thereby improving the processing efficiency of the topic generation device. (3) If the identification material is a static material object material, the topic generation device can identify the static material object by calling an interface for identifying the material object, and it can be understood that the topic generation device can identify the background in the multimedia data based on the mode, and further can determine a multimedia fragment including a certain background. (4) If the material is identified to inquire the text material (such as names, alias names and fan names of main actors), the topic generation equipment can perform text identification on the text material by calling a natural language identification model, so as to determine a multimedia segment including a certain text characteristic.

In another embodiment, when the number of the one or more pieces of target comment content is one, and the piece of target comment content is one multimedia segment in the multimedia data, the topic generation device may directly compare the multimedia segment with the multimedia data, determine the position of the multimedia segment in the multimedia data, and further take the determined position as the starting position of the target multimedia segment matching the discussion topic.

Based on the above description, the specific process of the topic generation device determining the target multimedia segment matching the discussion topic may exemplarily refer to fig. 7a, wherein the topic generation device may exemplarily be: the server cluster is composed of a multimedia data understanding server and a semantic understanding server. As shown in fig. 7a, an operator may first upload multimedia data of a discussion topic and identification materials that may be needed in a topic generation process to a multimedia data understanding server; it is understood that, since one identification material may have a plurality of features, for example, when the identification material uploaded by an operator is a video clip, the video clip may include human faces and character actions; in this case, the video segment is simply uploaded as the identification material, and the multimedia data understanding server does not know whether the video segment is used for extracting the content corresponding to the face feature or the content corresponding to the action feature in the video a, and therefore it is necessary to arbitrarily determine and upload the feature to be identified of the identification material, and determine whether the identification material is specifically the face material or the action material. Next, the multimedia data understanding server may perform feature detection based on the identification material and the multimedia data, and store the time point of occurrence of the feature corresponding to the identification material and the feature in association (e.g., in a memory of the multimedia data understanding server). Further, after the multimedia data understanding server receives the features of the keywords of the discussion topic sent by the semantic understanding server, the multimedia data understanding server can directly search the corresponding multimedia fragments from the memory of the multimedia data understanding server based on the features of the keywords, and then obtain the target multimedia fragments conforming to the features of all the keywords.

Based on the related description of fig. 7a, it can be understood that, when the target comment content is the bullet screen content, the specific process of the topic generation device determining the target multimedia segment may exemplarily refer to the process shown in fig. 7 b; when the target comment content is the comment content of the comment area, a specific flow of the topic generation device determining the target multimedia segment may be as shown in fig. 7c or fig. 7 d. Fig. 7b the trending topic server, the barrage server, the semantic understanding server, and the video understanding server may be integrated in one topic generation device, which includes the functions of each of the trending topic server, the barrage server, the semantic understanding server, and the video understanding server; the trending topic server, the comment server, the semantic understanding server, and the video understanding server in fig. 7c may be integrated in one topic generation device that includes the functions of each of the trending topic server, the comment server, the semantic understanding server, and the video understanding server. Since the specific implementation principle of each step in the flowcharts shown in fig. 7b and fig. 7c is consistent with the description in fig. 7a, and the specific implementation principle of each step in the flowchart shown in fig. 7d is consistent with the related step in fig. 3 or fig. 7a, the detailed description of the present application is omitted here.

S502, determining the initial position of the target multimedia segment in the multimedia data.

Wherein the topic generation device may take a time when the features of all keywords of the discussion topic simultaneously appear for the first time as a starting position of the target multimedia segment in the multimedia data.

S503, establishing an association relationship between the discussion topic and the starting position, and storing the discussion topic and the starting position associated with the discussion topic into a memory.

In an embodiment, the related description of step S503 may refer to step S406, which is not described herein again.

It is understood that the generated discussion thread may be displayed in the playing interface where the multimedia data is located, or may be displayed in the discussion interface, as shown in fig. 8, the discussion interface 81 may include one or more discussion threads, and the user may view a plurality of discussion threads in the discussion interface 81, for example: the user can trigger the display of the topic detail page 82 by clicking on the display area 811 of the discussion topic on the terminal device, and the topic detail page 82 includes multimedia data 821 (assuming that the multimedia data 821 is video data) corresponding to the discussion topic 812 and one or more pieces of comment content 822; it should be noted that, in a specific implementation, when the terminal device detects a multimedia clip playing instruction for the discussion topic 812 (for example, the terminal device detects that the user clicks the playing entry 8211), a multimedia clip playing request may be generated and sent to the topic generation device, where the multimedia clip playing request may carry a topic identifier of the discussion topic; for example: the terminal device may generate and send a multimedia segment playing request carrying a topic identifier of the discussion topic 812 (such as the name "XXXX 4" of the discussion topic) to the topic generation device, so that the topic generation device may respond to the multimedia segment playing request, search a starting position associated with the discussion topic corresponding to the topic identifier in a memory, and then obtain a target multimedia segment matching the discussion topic, where the starting position of the target multimedia segment may be the starting position obtained by the topic generation device in the memory; further, the topic generation device may transmit the obtained target multimedia segment to the terminal device, so that the terminal device may directly start playing from the start position of the target multimedia segment in the multimedia data 821. Optionally, the topic generation device may further obtain an end position of the target multimedia segment, based on which the terminal device may stop playing the multimedia data after playing to the end position of the target multimedia segment, or continue playing the multimedia data after playing to the end position of the target multimedia segment, which is not limited in this application.

According to the embodiment of the application, the target multimedia segment and the discussion topic are stored in a correlated mode, so that the user can directly jump to the target multimedia segment which is strongly related to the discussion topic when participating in the discussion topic, and the experience of the user when participating in the topic discussion is improved to a certain extent.

Based on the description of the above embodiment of the topic generation method, the embodiment of the present application also discloses a topic generation apparatus, which may be a computer program (including program code) running in the above mentioned server. The topic generation apparatus may perform the methods illustrated in fig. 1, 4 or 5. Referring to fig. 9, the topic generation apparatus 90 may include at least: semantic recognition section 901 and generation section 902.

A semantic recognition unit 901, configured to perform semantic recognition on one or more pieces of target comment content of the multimedia data to obtain target semantic information; wherein, the topicality corresponding to the target comment content or the target semantic information meets a preset topicality condition, and the topicality is used for indicating the participation degree of the multimedia data;

a generating unit 902, configured to generate a discussion topic of the multimedia data based on the target semantic information.

In an embodiment, the semantic identifying unit 901 may be specifically configured to:

performing semantic recognition on the one or more pieces of target comment content to obtain one or more pieces of semantic information;

acquiring the number of comment contents corresponding to similar semantic information in the one or more pieces of semantic information;

if the number of the comment contents corresponding to the target similar semantic information meets a first preset number condition, determining that the topicality corresponding to the target similar semantic information meets a preset topicality condition;

and taking the target similar semantic information as the target semantic information.

In another embodiment, the semantic identifying unit 901 may be specifically configured to:

acquiring interactive information of each piece of comment content in one or more pieces of comment content of the multimedia data;

determining the corresponding topicality of each piece of comment content according to the interactive information of each piece of comment content;

determining target comment contents with topicality meeting the preset topicality condition in the comment contents;

performing semantic identification on the target comment content to obtain semantic information of the target comment content;

and taking the semantic information of the target comment content as the target semantic information.

In yet another embodiment, the topic generation apparatus further comprises a processing unit 903, the processing unit 903 is configured to:

acquiring the playing time of the multimedia data;

dividing the playing time into a plurality of playing time periods;

acquiring the number of segment comment contents of multimedia segments played in each playing time period in the plurality of playing time periods, wherein the multimedia segments played in each playing time period are the multimedia segments of the multimedia data, and the semantic information of the segment comment contents is the target semantic information;

determining the number of the segment comment contents meeting a second preset number condition, and determining the initial position of a target multimedia segment matched with the discussion topic in the multimedia data according to the time information of the segment comment content with the earliest release time in the determined segment comment contents;

and establishing an association relationship between the discussion topic and the starting position, and storing the discussion topic and the starting position associated with the discussion topic into a memory.

In yet another embodiment, the processing unit 903 is further configured to:

determining a target multimedia segment matching the discussion topic in the multimedia data;

determining a starting position of the target multimedia segment in the multimedia data;

In yet another embodiment, the processing unit 903 may further be configured to:

receiving a multimedia segment playing request sent by a terminal device, wherein the multimedia segment playing request carries a topic identifier of the discussion topic, and the multimedia segment playing request is sent by the terminal device when a multimedia segment playing instruction for the discussion topic is detected;

searching a starting position associated with the discussion topic corresponding to the topic identification in the memory;

acquiring a target multimedia segment matched with the discussion topic, wherein the initial position of the target multimedia segment in the multimedia data corresponding to the discussion topic is a searched initial position;

and sending the target multimedia segment to the terminal equipment so as to enable the terminal equipment to play the target multimedia segment.

In yet another embodiment, the processing unit 903 may be further configured to:

acquiring the playing time of the multimedia data;

dividing the playing time into a plurality of playing time periods;

determining first multimedia fragments corresponding to the fragment comment contents of which the number meets a second preset number condition;

determining a second multimedia segment adjacent to the first multimedia segment in the multimedia data, wherein the second multimedia segment is matched with the discussion topic corresponding to the first multimedia segment; the playing time of the second multimedia segment is earlier than that of the first multimedia segment, or the playing time of the second multimedia segment is later than that of the first multimedia segment, or the playing time of a first multimedia sub-segment in the second multimedia segment is earlier than that of the first multimedia segment, and the playing time of a second multimedia sub-segment in the second multimedia segment is later than that of the first multimedia segment;

and obtaining the target multimedia fragment based on the first multimedia fragment and the second multimedia fragment.

In another embodiment, the processing unit 903, when acquiring the number of the segment comment contents of the multimedia segment played in each of the plurality of playing time periods, may be configured to:

acquiring time information of each piece of target comment content in the one or more pieces of target comment content in the multimedia data playing process;

determining the segment comment content of the multimedia segment played in each playing time period according to the time information of each piece of target comment content and each playing time period, wherein the segment comment content is part or all of the one or more pieces of target comment content;

and acquiring the number of the segment comment contents of the multimedia segments played in each playing time period.

According to an embodiment of the present application, the steps involved in the methods shown in fig. 1 and fig. 4 and fig. 5 may be performed by the units in the topic generation apparatus 90 shown in fig. 9. For example, step S101 shown in fig. 1 may be performed by the semantic recognition unit 901 in the topic generation apparatus 90 shown in fig. 9, and step S102 may be performed by the generation unit 902 in the topic generation apparatus 90 shown in fig. 9. As another example, steps S401 to S406 shown in fig. 4 may all be performed by the processing unit 903 in the topic generation apparatus 90 shown in fig. 9; for another example, steps S501 to S503 shown in fig. 5 may be performed by the processing unit 903 in the topic generation apparatus 90 shown in fig. 9.

According to another embodiment of the present application, the units in the topic generation apparatus 90 shown in fig. 9 are divided based on logical functions, and the units may be respectively or totally combined into one or several other units to form, or some unit(s) may be further split into multiple units with smaller functions to form, which may achieve the same operation without affecting the achievement of the technical effects of the embodiments of the present application. In other embodiments of the present application, the topic generation device may also include other units, and in practical applications, these functions may also be implemented by being assisted by other units, and may be implemented by cooperation of multiple units.

According to another embodiment of the present application, the topic generation apparatus 90 as shown in fig. 9 and the topic generation method of the embodiment of the present application may be configured and implemented by running a computer program (including program code) capable of executing the steps involved in the method as shown in fig. 1 or fig. 4 or fig. 5 on a general-purpose computing device such as a computer including a processing element and a storage element such as a Central Processing Unit (CPU), a random access storage medium (RAM), a read-only storage medium (ROM), and the like. The computer program may be embodied on a computer storage medium, for example, and loaded into and executed by the computer apparatus described above via the computer storage medium.

The topic generation device provided by the embodiment of the application carries out semantic identification on one or more target comment contents of the multimedia data to obtain the target semantic information, and then generates the discussion topic of the multimedia data based on the target semantic information. In addition, as the exclusive discussion topic is generated for the discussion hotspot, the multimedia data can have more topic quantity.

Based on the description of the above method embodiment and apparatus embodiment, the present application embodiment further provides a device for generating a topic, please refer to fig. 10, the topic generating device 100 at least includes a processor 1001 and a computer storage medium 1002, and the processor 1001 and the computer storage medium 1002 may be connected by a bus or other means.

The computer storage medium 1002 is a memory device in a computer device for storing programs and data. It is understood that the computer storage medium 1002 herein may include a built-in storage medium in the topic generation device, and may also include an extended storage medium supported by the topic generation device. The computer storage media 1002 provides storage space that stores the operating system of the topic generation apparatus. Also stored in this memory space are one or more instructions, which may be one or more computer programs (including program code), suitable for loading and execution by processor 1001. The computer storage medium may be a high-speed RAM memory, or may be a non-volatile memory (non-volatile memory), such as at least one disk memory; and optionally at least one computer storage medium located remotely from the processor. The processor 1001 (or CPU) is a computing core and a control core of the topic generation device, and is adapted to implement one or more instructions, and specifically, is adapted to load and execute one or more instructions so as to implement a corresponding method flow or a corresponding function.

In one embodiment, one or more instructions stored in the computer storage medium 1002 may be loaded and executed by the processor 1001 to implement the corresponding method steps described above in connection with the method embodiments illustrated in fig. 1 and 4 and 5; in particular implementations, one or more instructions in the computer storage medium 1002 are loaded by the processor 1001 and perform the following steps:

In one embodiment, the processor 1001 may be specifically configured to:

In yet another embodiment, the processor 1001 may be further configured to perform:

In yet another embodiment, the processor 1001 may be specifically configured to:

acquiring the playing time of the multimedia data;

dividing the playing time into a plurality of playing time periods;

acquiring the playing time of the multimedia data;

dividing the playing time into a plurality of playing time periods;

The topic generation device provided by the embodiment of the application carries out semantic identification on one or more pieces of target comment content of the multimedia data to obtain the target semantic information, and then generates the discussion topic of the multimedia data based on the target semantic information. In addition, as the exclusive discussion topic is generated for the discussion hotspot, the multimedia data can have more topic quantity.

The embodiment of the present application further provides a computer storage medium, where a computer program of the topic generation method is stored in the computer storage medium, where the computer program includes program instructions, and when one or more processors load and execute the program instructions, the description of the topic generation method in the embodiment may be implemented, which is not described herein again. The description of the beneficial effects of the same method is not repeated herein. It will be understood that the program instructions may be deployed to be executed on one or more devices capable of communicating with each other.

It should be noted that according to an aspect of the present application, a computer program product or a computer program is also provided, and the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer readable storage medium. The processor in the topic generation device reads the computer instructions from the computer readable storage medium and then executes the computer instructions, thereby enabling the topic generation device to perform the methods provided in the various alternatives in the aspect of the embodiments of topic generation methods illustrated in fig. 1 and 4 and 5 described above.

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

While the invention has been described with reference to a particular embodiment, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. A topic generation method, comprising:

2. The method of claim 1, wherein the semantically recognizing one or more pieces of target comment content of the multimedia data to obtain target semantic information comprises:

3. The method of claim 1, wherein the semantically recognizing one or more pieces of target comment content of the multimedia data to obtain target semantic information comprises:

4. The method of claim 1, further comprising:

acquiring the playing time of the multimedia data;

dividing the playing time into a plurality of playing time periods;

5. The method of claim 1, further comprising:

6. The method according to claim 4 or 5, characterized in that the method further comprises:

7. The method of claim 5, wherein the determining the target multimedia segment in the multimedia data that matches the discussion thread comprises:

acquiring the playing time of the multimedia data;

dividing the playing time into a plurality of playing time periods;

8. The method according to claim 4 or 7, wherein the obtaining of the number of the segment comment contents of the multimedia segment played in each of the plurality of playing time periods comprises:

9. A topic generation apparatus, comprising:

10. A computer storage medium having stored thereon one or more instructions adapted to be loaded by the processor and to perform the topic generation method of any of claims 1-8.