Video title generation method and apparatus, electronic device, storage medium, and product thereof

Abstract

This disclosure provides a method, apparatus, electronic device, storage medium, and products for generating video titles, relating to the field of big data processing technology, and particularly to the field of video data processing technology. The specific implementation of the video title generation method is as follows: extracting main words and feature words from the video information of the video to be processed; generating initial tags based on the main words and feature words; the initial tags containing at least one main word and at least one feature word; filtering target tags from the initial tags based on the search question of the video to be processed; and generating the title of the video to be processed based on the target tags.

Description

Technical Field

[0001] This disclosure relates to the field of data processing technology, and in particular to video title generation methods, apparatus, electronic devices, storage media and products thereof. Background Technology

[0002] As a crucial element of a video, the title should be highly relevant and traffic-generating. Currently, video titles are generally set by the video creators themselves. However, inexperienced video creators often struggle to produce high-quality titles, resulting in titles that fail to attract users and consequently, lower click-through rates and poor conversion rates. Summary of the Invention

[0003] This disclosure provides a video title generation method, apparatus, electronic device, storage medium, and product thereof.

[0004] According to a first aspect of this disclosure, a video title generation method is provided, comprising:

[0005] Extract the main words and the feature words of the main words from the video information of the video to be processed;

[0006] An initial label is generated based on the subject word and the feature word; the initial label contains at least one subject word and at least one feature word.

[0007] Target tags are selected from the initial tags based on the search question of the video to be processed;

[0008] Generate a title for the video to be processed based on the target tag.

[0009] According to a second aspect of this disclosure, a video title generation apparatus is provided, comprising:

[0010] The extraction module is used to extract the main words and the feature words of the main words from the video information of the video to be processed;

[0011] A tag generation module is used to generate initial tags based on the main word and the feature word; the initial tags contain at least one main word and at least one feature word;

[0012] The filtering module is used to filter target tags from the initial tags based on the search question of the video to be processed;

[0013] The title generation module is used to generate a title for the video to be processed based on the target tag.

[0014] According to a third aspect of this disclosure, an electronic device is provided, comprising:

[0015] At least one processor; and

[0016] A memory communicatively connected to the at least one processor; wherein,

[0017] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the video title generation methods described above.

[0018] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are configured to cause the computer to perform the video title generation method according to any of the preceding claims.

[0019] According to a fifth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the video title generation method according to any of the preceding claims.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0021] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0022] Figure 1 This is a flowchart illustrating a video title generation method according to an example embodiment of this disclosure;

[0023] Figure 2 This is a schematic diagram of the production logic of a video to be processed, as an example embodiment of this disclosure;

[0024] Figure 3 This is a flowchart of another video title generation method according to an example embodiment of this disclosure;

[0025] Figure 4 This is a flowchart illustrating the selection of target tags in a video title generation method according to an example embodiment of this disclosure;

[0026] Figure 5 This is a flowchart illustrating the filtering of target tags in another video title generation method according to an example embodiment of this disclosure;

[0027] Figure 6 This is a flowchart illustrating the filtering of target tags in another video title generation method according to an example embodiment of this disclosure;

[0028] Figure 7 This is a flowchart illustrating the filtering of target tags in another video title generation method according to an example embodiment of this disclosure;

[0029] Figure 8 This is a schematic diagram of another video title generation device according to an example embodiment of the present disclosure;

[0030] Figure 9 This is a block diagram of an electronic device used to implement the video title generation method of the embodiments of this disclosure. Detailed Implementation

[0031] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0032] Figure 1 This is a flowchart of a video title generation method according to an example embodiment of the present disclosure. The video title generation method includes the following steps:

[0033] Step 101: Extract the main words and feature words of the main words from the video information of the video to be processed.

[0034] The video to be processed can be a shot video, an animated video, or an AI (Artificial Intelligence) video after processing the shot video based on video enhancement technology; the video to be processed can be obtained locally or online; the embodiments of this disclosure do not particularly limit the type or acquisition method of the video to be processed.

[0035] See Figure 2 Taking AI videos as an example, a complete AI video may include, but is not limited to, a title, a cover image, and content. The content portion occupies a large portion of the AI ​​video and includes video footage, narration, subtitles, background music, transition animations, and voice-over. Therefore, the video information of the video to be processed may include, but is not limited to, at least one of the following: the existing title of the video to be processed, the text of the video to be processed, the subject information of the subject associated with the video to be processed, and the descriptive information of the video to be processed.

[0036] The aforementioned existing title can be a title set by the user for the video to be processed, or the title of the subject associated with the video to be processed can be determined as the existing title of the video to be processed. This disclosure does not impose any particular limitation on this.

[0037] The aforementioned text includes at least one of the following: subtitle text of the video to be processed, explanatory text and narration added to the video based on video enhancement technology, etc.

[0038] The subject mentioned above refers to the objects involved in the video. For example, for filmed videos and AI videos, the subject is the people, objects, or other subjects being filmed; for animated videos, the subject is the characters in the video.

[0039] Subject terms are used to represent the subject, such as the subject's name. Feature terms represent the subject's characteristics, such as the subject's model, size, color, application scenario, and purpose.

[0040] Step 102: Generate initial labels based on the main words and feature words.

[0041] The initial tags are obtained based on the combination of subject words and feature words. The number of initial tags is related to the number of subject words and feature words obtained in step 101. Each initial tag contains at least one subject word and at least one feature word. The number of subject words in each initial tag can be set according to the actual situation, and similarly, the number of feature words in each initial tag can also be set according to the actual situation. The number of subject words and feature words in each initial tag can be the same or different.

[0042] Step 103: Filter out target tags from the initial tags based on the search question of the video to be processed.

[0043] The so-called search question is the question of how to find the video to be processed based on the search question, and / or the question of how to find the video information of the video to be processed based on the search question.

[0044] In step 103, target tags are selected from the initial tags based on the user's search query. These target tags can largely reflect the user's needs and ensure that the final title has a certain traffic potential. The number of selected target tags may be one, two, or more.

[0045] Step 104: Generate the title of the video to be processed based on the target tags.

[0046] If the number of target tags selected in step 103 is 1, then that target tag is determined as the title of the video to be processed.

[0047] If at least two target tags are selected in step 103, these two target tags are combined, and the combined result is used as the title of the video to be processed. The order of the multiple target tags is determined according to a semantic algorithm to make the resulting title easy to understand.

[0048] In this embodiment of the disclosure, subject words and feature words are extracted from the video information of the video to be processed to generate initial tags. These initial tags are highly relevant to the video content. Then, target tags are selected from the initial tags based on the search question. These target tags can largely reflect user needs and have a certain traffic base. Thus, the title generated based on the target tags can express the main idea of ​​the video to be processed, guide user interest, and improve the click-through rate and output rate of the video to be processed.

[0049] The video title generation method provided in this disclosure is applicable to automatically generating titles for various platforms and videos. It is particularly suitable for generating titles for B2B (business-to-business communication via dedicated networks or the Internet) industry knowledge-based videos. Moreover, it does not require data support from industrial product knowledge graphs, and can generate titles that are highly relevant to the video content and have a traffic base, thus meeting the video's distribution performance requirements on search channels.

[0050] In one embodiment, step 101 includes: performing word segmentation on the video information, and extracting subject words and feature words from the result of the word segmentation. Specific algorithms for word segmentation can be found in related technologies and will not be elaborated here.

[0051] The video information includes at least one of the following: the existing title of the video to be processed, the text of the video to be processed, and the subject information of the subject associated with the video to be processed. Word segmentation is performed on the video information, that is, at least one of the existing title of the video to be processed, the text of the video to be processed, and the subject information of the subject associated with the video to be processed is segmented to obtain subject words and feature words that are highly relevant to the video content, thereby obtaining initial tags that are highly relevant to the video content.

[0052] In one embodiment, the results of word segmentation are classified using a classification model to determine the subject word and feature words. The network architecture and training process of the classification model can be found in related technologies and will not be elaborated here.

[0053] See Figure 3 The diagram illustrates an example where the main subject is a product, and the video information includes both video text and main subject information. The main subject information further includes product information and merchant information. Word segmentation of the video information typically yields main subject words, feature words, and semantically insensitive words. In scenarios where the main subject is a product, refer to Table 1. Main subject words refer to industrial product entity words expressing the product, such as: steamed bun machine, cutting machine, etc. Feature words include descriptive words and marketing words. Descriptive words describe the product's characteristics, effects, and parameters (excluding descriptions related to sales). Marketing words are words related to product sales, such as descriptions of product price, manufacturer, etc., such as: source manufacturer, compare prices, lowest price online, etc. Semantically insensitive words are words without actual meaning, used for the sake of fluency, such as conjunctions, and require filtering.

[0054] In scenarios where the main subject is a product, taking the video to be processed as an AI video as an example, the AI ​​video generates video text based on the statistical characteristics of product parameters and a specific video type (theme). Therefore, the video text and video type information can be obtained in advance. At the same time, the video material uses product videos of the category operated by the merchant. Therefore, the AI ​​video can obtain product information and merchant information by associating products.

[0055] Table 1

[0056]

[0057] In one embodiment, step 102 involves randomly combining the subject word and the feature word to obtain the initial label.

[0058] The following example illustrates how to generate initial tags, where each initial tag contains one subject word and one feature word: If step 101 yields two subject words and three feature words, namely subject word A, subject word B, feature word a, feature word b, and feature word c, then six initial tags can be obtained: subject word A feature word a, subject word A feature word b, subject word A feature word c, subject word B feature word a, subject word B feature word b, and subject word B feature word c.

[0059] It should be noted that the order of the main word and feature words in the initial tag is not limited to the above-described order of main word first and feature word last. The order of the main word and feature words can be determined according to semantics. For example, for the initial tag "high-strength bolt", the feature word "high-strength" comes first and the main word "bolt" comes second; for the initial tag "bolt price", the feature word "price" comes first and the main word "bolt" comes second. The order of the main word and feature words can also be determined according to the sentence structure corresponding to the original video information. Taking the existing title of the video to be processed, "High-strength bolts, self-produced and sold directly, complete specifications, passed tensile and hardness tests", as an example, the main word "bolt" and the feature word "high-strength" are extracted, with the feature word first and the main word last. When combining the main word and feature words, the order of feature word first and main word last is also followed to obtain the initial tag "high-strength bolt".

[0060] See Table 2, which illustrates the initial tags obtained by segmenting the product titles contained in the video information.

[0061] Table 2

[0062]

[0063] In one embodiment, to increase the number of tags, the subject words and / or feature words are replaced. Specifically, the subject words and feature words are combined to obtain an initial combination result. The feature words and / or subject words in the initial combination result are replaced with words of the same type to obtain a replaced combination result. Initial tags are generated based on the initial combination result and the combined result.

[0064] In one embodiment, when the number of initial combination results is less than a threshold, the feature words and / or subject words in the initial combination results are replaced with words of the same type. When the number of initial combination results is greater than or equal to the threshold, the initial combination results are determined as the initial labels.

[0065] In one implementation, the same type of words includes synonyms (or alternative names, common names). The same type of words are used to replace the feature words in the initial combination result, that is, the feature words in the initial combination result are replaced with synonyms of the feature words; the same type of words are used to replace the main words in the initial combination result, that is, the main words in the initial combination result are replaced with synonyms of the main words.

[0066] In one implementation, the types of similar words correspond to the types included in the feature words. Similar words can be obtained from a preset thesaurus. For example, the preset thesaurus contains two types of words: descriptive words and marketing words. If the feature words include descriptive words, then similar words are obtained from the descriptive words in the preset thesaurus to replace the feature words in the initial combination result; if the feature words include marketing words, then similar words are obtained from the marketing words in the preset thesaurus to replace the marketing words in the initial combination result.

[0067] In one implementation, the existing title contained in the video information is segmented to obtain the main word and the feature word. Similar words of the feature word are obtained from the text and / or main information contained in the video information, and the feature word in the initial combination result is replaced with similar words. The main word in the initial combination result is replaced with a synonym of the main word, or the main word is not replaced.

[0068] See Table 3, which shows the results after expanding the initial tags. When initial tags are obtained, they are replaced according to their structure, thus exponentially increasing the production volume of initial tags. The expansion logic of the initial tags involves replacing them according to the sentence structure and similar types of words.

[0069] Table 3

[0070]

[0071] In one embodiment, see Figure 4 Step 103 includes:

[0072] Step 103-1a: Perform semantic clustering on the search problem to obtain at least two first sets.

[0073] Semantic clustering of search queries aims to semantically map the search queries to the video text, thereby improving the matching range while ensuring a high relevance between the initial tags and the search queries. Each first set can contain one, two, or even more search queries.

[0074] Step 103-2a: Determine the total number of searches for the search questions included in each first set within the preset time period.

[0075] The duration, start time, and end time of the preset time period can be determined according to actual needs, and this embodiment does not impose any special limitations on them.

[0076] Taking the four search questions shown in Table 4 as examples, for the first set whose semantic type is "use", the total number of searches for each question in the first set within the preset time period is 110; for the first set whose semantic type is "price", the total number of searches for each question in the first set within the preset time period is 80.

[0077] Table 4

[0078] Search issues Search count Matched the main word Semantic types Excavator use 10 excavator Use class Excavator Operation Tutorial 100 excavator Use class Excavator Price Trends 80 excavator Price category dough mixer 50 -(No match found) -

[0079] Step 103-3a: Identify the feature words that match the semantics of the search questions contained in the first target set as target feature words.

[0080] The first target set is the first set with the highest total number of searches.

[0081] Taking Table 4 as an example, the first set with the semantic type "use" has the highest total number of searches. Therefore, the feature words that match "use" are identified as target feature words. Target feature words include "use", "operation tutorial" and related or similar words.

[0082] Step 103-4a: Determine the initial label containing the target feature word as the target label.

[0083] In this embodiment, by performing semantic clustering on the search question and filtering target tags based on the clustering results, the matching range of feature words can be improved, and target tags with high relevance to the video to be processed can be obtained, ensuring that the final video title has a certain traffic base and strong relevance.

[0084] The number of initial labels containing the target feature words in step 103-4a can be one, two, or even more. When the number of initial labels containing the target feature words is large, the resulting title is longer and can be further filtered to refine it.

[0085] When the number of initial labels containing the target feature words is large, the following is a method for further filtering the initial titles. See [link to relevant documentation]. Figure 5 Step 103-4a includes the following steps:

[0086] Step 103-41a: If the number of initial labels containing the target feature words is greater than the number threshold, then count the occurrence times of each target feature word.

[0087] The quantity threshold can be determined according to the actual situation, for example, it can be 2.

[0088] The frequency of occurrence includes at least one of the following: the frequency of occurrence of the target feature word in the text of the video to be processed, the frequency of occurrence of words of the same type as the target feature word in the text of the video to be processed, and the frequency of occurrence of feature words with different meanings from the target feature word in the text of the video to be processed.

[0089] Different meanings mean opposite meanings to the target feature word, different parameter values, etc. For example, for the same subject, the target feature word A is "small", while the feature word in the video text is "large", which means different things; the feature value of the target feature word B is "mechanical strength 38N", while the feature word in the video text is "mechanical strength 20N", which means different things.

[0090] Step 103-42a: Determine the relevance between the initial labels containing the target feature words and the video to be processed based on the frequency of occurrence.

[0091] Theoretically, there is no difference in relevance. However, since the goal is to match the title with the content type of the video to be processed in terms of user experience, the relevance of the initial label to the video to be processed can be determined based on the frequency of occurrence of the target feature words in the video text.

[0092] Taking the determination of relevance based on the frequency of occurrence of the target feature word in the text of the video to be processed and the frequency of occurrence of feature words with different meanings from the target feature word in the text of the video to be processed as an example, the formula for calculating the relevance 'a' of the initial label can be, but is not limited to, expressed as follows:

[0093] a = 1 + (x - 3y) / (x + y + z);

[0094] x represents the number of times the target feature word appears in the text of the video to be processed; y represents the number of times feature words with different meanings from the target feature word appear in the text of the video to be processed; z represents the number of target feature words that do not appear in the text of the video to be processed. The above formula can be obtained by fitting historical data.

[0095] Step 103-43a: Determine the target feature words with a correlation greater than the correlation threshold as strongly correlated feature words, and determine the initial label containing strongly correlated feature words as the target label.

[0096] The correlation threshold can be set according to the actual situation.

[0097] In one embodiment, step 103-43a is replaced by step 103-43a'. Step 103-43a' involves sorting the relevance from largest to smallest and identifying at least one target feature word at the top of the sort as a strongly relevant feature word, and identifying the initial label containing the strongly relevant feature word as the target label.

[0098] In this embodiment, the initial tags are further filtered through video text to obtain target tags that are strongly related to the video to be processed. The matching degree between the title generated based on the target tags and the content type of the video to be processed is further improved, reducing the situation where the title and the content of the video to be processed do not match, improving the quality and readability of the title, and avoiding misleading clicks.

[0099] In one embodiment, see Figure 6 and Figure 7 Step 103 includes:

[0100] Step 103-1b: Perform subject clustering on the search question to obtain at least two second sets.

[0101] The subject and main word of the second target set are matched.

[0102] Subject clustering of search queries aims to map the subject of the search query to the corresponding subject terms, eliminating search queries where the subject and subject terms do not match. This mapping improves the matching range of the subject terms in the initial tags. Each second set can contain one, two, or even more search queries.

[0103] Step 103-2b: Perform semantic clustering on the search questions contained in the second target set to obtain at least two first sets.

[0104] Step 103-3b: Determine the total number of searches for the search questions included in each first set within the preset time period.

[0105] Step 103-4b: Identify the feature words that match the semantics of the search questions contained in the first target set as target feature words.

[0106] Among them, the first target set is the first set with the highest total number of searches;

[0107] Step 103-5b: Determine the initial label containing the target feature word as the target label.

[0108] The specific implementation methods of steps 103-2b to 103-5b are similar to those of steps 103-1a to 103-4a, and will not be repeated here.

[0109] In this embodiment, by performing subject-based and semantic clustering on the search question and filtering target tags based on the clustering results, the matching range of feature words can be improved, and target tags with strong relevance to the video to be processed can be obtained, ensuring that the final video title has a certain traffic base and strong relevance.

[0110] In one embodiment, after the target tags are determined, they are combined according to a semantic algorithm to obtain the title of the video to be processed, resulting in a more readable and easier-to-understand title. The semantic algorithm can be described using relevant technologies, which will not be elaborated here.

[0111] In one embodiment, after determining the target label, the target label is input into a semantic model, and the semantic model outputs the title of the video to be processed, thus obtaining a more readable and easier-to-understand title. The semantic model is trained from labeled samples of video titles, and the label samples contain subject words and feature words. For the specific training process of the semantic model, please refer to the relevant technical descriptions, which will not be repeated here.

[0112] In one embodiment, target tags are combined according to a semantic algorithm to obtain a combination result. An effect sentence is added to the combination result of the target tags, and the combination result with the effect sentence is determined as the title of the video to be processed.

[0113] In one embodiment, the target label is input into the semantic model to obtain a combined result. An effect sentence is added to the combined result of the target label, and the combined result with the effect sentence is determined as the title of the video to be processed.

[0114] Effect sentences can be sentences related to materials and main features, set according to user needs. Effect sentences can be randomly selected from a pre-set sentence library, such as sentences like "An experienced master will guide you through the current state of the industry" and "True masters are among the people," etc. Adding effect sentences can improve the interest of the title.

[0115] Corresponding to the aforementioned video title generation method embodiments, this disclosure also provides embodiments of a video title generation apparatus.

[0116] Figure 8 This disclosure provides a schematic diagram of a video title generation apparatus as an exemplary embodiment, the apparatus comprising:

[0117] Extraction module 81 is used to extract the main words and the feature words of the main words from the video information of the video to be processed;

[0118] The tag generation module 82 is used to generate initial tags based on the main words and the feature words; the initial tags contain at least one main word and at least one feature word;

[0119] The filtering module 83 is used to filter target tags from the initial tags according to the search question of the video to be processed;

[0120] The title generation module 84 is used to generate a title for the video to be processed based on the target tag.

[0121] Optionally, the video information includes at least one of the following: an existing title of the video to be processed, the text of the video to be processed, and subject information of the subject associated with the video to be processed;

[0122] The extraction module includes:

[0123] The word segmentation unit is used to segment the video information into words.

[0124] An extraction unit is used to extract the main word and the feature word from the result of the word segmentation process.

[0125] Optionally, the label generation module includes:

[0126] A combination unit is used to combine the main word and the feature word to obtain an initial combination result;

[0127] The replacement unit is used to replace the feature words in the initial combination result with words of the same type and / or replace the main words in the initial combination result with words of the same type to obtain the replacement combination result;

[0128] A generation unit is used to generate the initial label based on the initial combination result and the initial combination result.

[0129] Optionally, the filtering module includes:

[0130] Clustering unit, used to perform semantic clustering on the search problem to obtain at least two first sets;

[0131] The number of searches determination unit is used to determine the total number of searches for the search questions contained in each first set within a preset time period;

[0132] A word determination unit is used to determine target feature words as feature words whose semantics match the search question contained in the first target set; wherein, the first target set is the first set with the highest total number of searches;

[0133] A label determination unit is used to determine the initial label containing the target feature word as the target label.

[0134] Optionally, the clustering unit is specifically used for:

[0135] Subject clustering is performed on the search problem to obtain at least two second sets;

[0136] Perform semantic clustering on the search questions contained in the second target set; wherein the subject corresponding to the second target set matches the main word.

[0137] Optionally, the label determining unit is specifically used for:

[0138] If the number of initial tags containing the target feature word is greater than the number threshold, the occurrence count of each target feature word is counted. The occurrence count includes at least one of the following: the occurrence count of the target feature word in the text of the video to be processed, the occurrence count of words of the same type as the target feature word in the text of the video to be processed, and the occurrence count of feature words with different meanings from the target feature word in the text of the video to be processed.

[0139] The relevance between the initial label containing the target feature word and the video to be processed is determined based on the frequency of occurrence.

[0140] Target feature words whose relevance is greater than the relevance threshold are identified as strongly related feature words, or the relevance is sorted from largest to smallest and at least one of the top-ranked target feature words is identified as strongly related feature words.

[0141] The initial label containing the strongly related feature words is determined as the target label.

[0142] Optionally, the title generation module is specifically used for:

[0143] The title of the video to be processed is obtained by combining the target tags according to the semantic algorithm;

[0144] Alternatively, the target label can be input into a semantic model, and the title of the video to be processed can be output based on the semantic model; wherein the semantic model is trained from label samples labeled with video titles, and the label samples contain subject words and feature words.

[0145] Optionally, the title generation module is specifically used for:

[0146] In the combined result of the target tags, an effect sentence is added so that the combined result with the added effect sentence is determined as the title of the video to be processed.

[0147] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0148] The technical solutions disclosed herein involve the collection, storage, use, processing, transmission, provision, and disclosure of the videos to be processed, all of which comply with relevant laws and regulations and do not violate public order and good morals.

[0149] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0150] Figure 9 A schematic block diagram of an example electronic device 900 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0151] like Figure 9 As shown, device 900 includes a computing unit 901, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 902 or a computer program loaded into random access memory (RAM) 903 from storage unit 908. RAM 903 may also store various programs and data required for the operation of device 900. The computing unit 901, ROM 902, and RAM 903 are interconnected via bus 904. Input / output (I / O) interface 905 is also connected to bus 904.

[0152] Multiple components in device 900 are connected to I / O interface 905, including: input unit 906, such as keyboard, mouse, etc.; output unit 907, such as various types of monitors, speakers, etc.; storage unit 908, such as disk, optical disk, etc.; and communication unit 909, such as network card, modem, wireless transceiver, etc. Communication unit 909 allows device 900 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0153] The computing unit 901 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 901 performs the various methods and processes described above, such as the video title generation method. For example, in some embodiments, the video title generation method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on device 900 via ROM 902 and / or communication unit 909. When the computer program is loaded into RAM 903 and executed by the computing unit 901, one or more steps of the video title generation method described above may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured to perform the video title generation method by any other suitable means (e.g., by means of firmware).

[0154] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0155] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0156] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0157] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0158] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0159] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0160] The computer-readable storage medium provided in this disclosure is a non-transitory computer-readable storage medium with computer instructions, wherein the computer instructions are used to cause the computer to perform the method provided in any of the above embodiments.

[0161] The computer program product provided in this disclosure includes a computer program that, when executed by a processor, implements the method provided in any of the above embodiments.

[0162] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0163] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for generating video titles, comprising: Extract the main words and the feature words of the main words from the video information of the video to be processed; The subject words are used to represent the subject, which is the object involved in the video, and the feature words represent the subject features; Initial labels are generated based on the main words and the feature words; The initial label contains at least one subject word and at least one feature word; The search problem for the video to be processed is subjected to semantic clustering to obtain at least two first sets; Determine the total number of searches for the search questions included in each first set within a preset time period; Feature words that match the semantics of the search questions contained in the first target set are identified as target feature words; wherein, the first target set is the first set with the highest total number of searches; If the number of initial tags containing the target feature word is greater than the number threshold, the occurrence count of each target feature word is counted. The occurrence count includes at least one of the following: the occurrence count of the target feature word in the text of the video to be processed, the occurrence count of words of the same type as the target feature word in the text of the video to be processed, and the occurrence count of feature words with different meanings from the target feature word in the text of the video to be processed. The relevance between the initial label containing the target feature word and the video to be processed is determined based on the frequency of occurrence. Target feature words whose relevance is greater than the relevance threshold are identified as strongly related feature words, or the relevance is sorted from largest to smallest and at least one target feature word at the top of the sort is identified as a strongly related feature word. The initial label containing the strongly related feature words is determined as the target label; Generate a title for the video to be processed based on the target tag.

2. The video title generation method according to claim 1, wherein the video information includes at least one of the following: an existing title of the video to be processed, the text of the video to be processed, and subject information of a subject associated with the video to be processed; Extract the main words and feature words of the main words from the video information of the video to be processed, including: The video information is segmented into words. The main word and the feature word are extracted from the result of the word segmentation process.

3. The video title generation method according to claim 1, comprising generating initial tags based on the main words and the feature words, including: The main words and the feature words are combined to obtain the initial combination result; Replace the feature words in the initial combination result with words of the same type and / or replace the main words in the initial combination result with words of the same type to obtain the replacement combination result; The initial label is generated based on the initial combination result and the replacement combination result.

4. The video title generation method according to claim 1, wherein performing semantic clustering on the search question includes: Subject clustering is performed on the search problem to obtain at least two second sets; Perform semantic clustering on the search questions contained in the second target set; wherein the subject corresponding to the second target set matches the main word.

5. The video title generation method according to any one of claims 1-4, wherein generating the title of the video to be processed based on the target tag comprises: The title of the video to be processed is obtained by combining the target tags according to the semantic algorithm; Alternatively, the target label can be input into a semantic model, and the title of the video to be processed can be output based on the semantic model; wherein the semantic model is trained from label samples labeled with video titles, and the label samples contain subject words and feature words.

6. The video title generation method according to claim 5 further includes: Add an effect sentence to the combined result of the target tags; The combined result of sentences with added effects is determined as the title of the video to be processed.

7. A video title generation device, comprising: The extraction module is used to extract the main words and the feature words of the main words from the video information of the video to be processed; The subject words are used to represent the subject, which is the object involved in the video, and the feature words represent the subject features; The tag generation module is used to generate initial tags based on the main words and the feature words; The initial label contains at least one subject word and at least one feature word; The filtering module is used to filter target tags from the initial tags based on the search question of the video to be processed; The title generation module is used to generate a title for the video to be processed based on the target tag; The filtering module includes: Clustering unit, used to perform semantic clustering on the search problem to obtain at least two first sets; The number of searches determination unit is used to determine the total number of searches for the search questions contained in each first set within a preset time period; A word determination unit is used to determine target feature words as feature words whose semantics match the search question contained in the first target set; wherein, the first target set is the first set with the highest total number of searches; The label determination unit is specifically used for: If the number of initial tags containing the target feature word is greater than the number threshold, the occurrence count of each target feature word is counted. The occurrence count includes at least one of the following: the occurrence count of the target feature word in the text of the video to be processed, the occurrence count of words of the same type as the target feature word in the text of the video to be processed, and the occurrence count of feature words with different meanings from the target feature word in the text of the video to be processed. The relevance between the initial label containing the target feature word and the video to be processed is determined based on the frequency of occurrence. Target feature words whose relevance is greater than the relevance threshold are identified as strongly related feature words, or the relevance is sorted from largest to smallest and at least one target feature word at the top of the sort is identified as a strongly related feature word. The initial label containing the strongly related feature words is determined as the target label.

8. The video title generation apparatus according to claim 7, wherein the video information includes at least one of the following: an existing title of the video to be processed, the text of the video to be processed, and subject information of a subject associated with the video to be processed; The extraction module includes: The word segmentation unit is used to segment the video information into words. An extraction unit is used to extract the main word and the feature word from the result of the word segmentation process.

9. The video title generation device according to claim 7, wherein the tag generation module comprises: A combination unit is used to combine the main word and the feature word to obtain an initial combination result; The replacement unit is used to replace the feature words in the initial combination result with words of the same type and / or replace the main words in the initial combination result with words of the same type to obtain the replacement combination result; A generation unit is used to generate the initial label based on the initial combination result and the replacement combination result.

10. The video title generation apparatus according to claim 9, wherein the clustering unit is specifically used for: Subject clustering is performed on the search problem to obtain at least two second sets; performing a same semantic clustering on search problems contained in the second target set; wherein The subject corresponding to the second target set matches the main word.

11. The video title generation apparatus according to any one of claims 7-10, wherein the title generation module is specifically used for: The title of the video to be processed is obtained by combining the target tags according to the semantic algorithm; Alternatively, the target tag can be input into a semantic model, and the title of the video to be processed can be output based on the semantic model; wherein, The semantic model is trained using labeled samples with video titles, and the labeled samples contain subject words and feature words.

12. The video title generation device according to claim 11, wherein the title generation module is specifically used for: In the combined result of the target tags, an effect sentence is added so that the combined result with the added effect sentence is determined as the title of the video to be processed.

13. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the video title generation method according to any one of claims 1-6.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the video title generation method according to any one of claims 1-6.

15. A computer program product comprising a computer program that, when executed by a processor, implements the video title generation method according to any one of claims 1-6.

Images