[0051] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.
[0052] Please refer to the attached figure 1 , is a flow chart of a method for generating a default search word disclosed in the present invention, such as figure 1 As shown, the method may include:
[0053] Step 101: Determine search word metadata according to global search history information;
[0054] According to the search history information of users on the whole network, the search word metadata can be determined. In this embodiment, the metadata may include channel popular search terms, frequent search terms and/or related search terms.
[0055] Wherein, the acquisition of popular search words of the channel may be based on the overall user search history, estimating the popular search words under each channel in the past day or several days, subject to the global search times, and each channel can retain at most the search times Top 10-20 popular search terms. Of course, the number of reserved popular search words is not fixed, and can be set by the manager according to the actual situation according to the actual application.
[0056] The determination of the frequent search term may be determined according to the frequency of the search term being searched by the user. Search terms can be classified into frequent search terms and infrequent search terms. For example, search terms such as TV series titles, variety show titles, and hot events, through behaviors such as chasing dramas and gossip, users have a high probability of frequent searches within a period of time, that is, frequent search terms; and search terms such as movie titles and long-tail search terms , after the user searches and clicks to watch, the probability of searching again in a short period of time is small, which is an infrequent search term.
[0057] Specifically, determining whether a search word is a frequent search word can be determined by a preset algorithm judging whether the result satisfies a condition. For example, based on the global user search history, the number of clicks and the number of clicks of each word in the past 15 days are counted. When the number of clicks on a search word is greater than 3000, and the number of clicks from the same user divided by the total number of clicks is greater than 0.36, That is, it is determined to be a frequent search term, otherwise it is an infrequent search term. Wherein, the same user refers to a user who clicks on the search term at least twice. For example, "Open the door" was clicked 10,000 times by users across the network, of which 2,000 users each clicked twice, and each of the remaining users clicked once. Then the total number of clicks from the same user is 2000*2=4000, and the number of clicks from different users is 10000-4000=6000, then "dividing all clicks from the same user by the total number of clicks" is 4000/10000= 0.4. If 0.4 is greater than 0.36, the word "open door" is a frequently searched word.
[0058] The related search words can adopt item-based collaborative filtering algorithm, take the search words of users in the whole network in the past 15 days as input, predict the relevant probability of each word, and take several words with the highest relevant probability as the search words related words.
[0059] Step 102: Determine the target user's preference data according to the target user's video history information;
[0060] Wherein, the preference data of the target user may be determined based on two aspects of video viewing history data and user search history data.
[0061] Each video has channel and label information, and according to the user's viewing video history data and preset rules, the target user's channel preference data and label preference data can be determined. Specifically, the user's video viewing history data in the past 30 days can be counted, and the number of views of each channel and label can be divided by the total number of views to obtain the score of the channel and label, that is, the target user's viewing preferences for channels and labels can be classified into Normalized to the probability of [0,1], the channel and label with probability greater than 0.4 can be set as the target user's preferred channel and preferred label.
[0062] Using the user's search history data, it is possible to determine the search terms that the user has searched for, and determine which search terms the user is interested in.
[0063] Step 103: Generate a plurality of default search terms according to the search term metadata and the preference data.
[0064] The determination of the default search term can combine the channel preference information and label preference information of the target user determined based on the target user’s video history information with the channel popular search terms, frequent search terms and related search terms determined based on the global search history information and search history data to determine. Based on the search situation of the entire network, the default search terms generated in combination with the personal hobbies of the target users are both public and pertinent, and it is easy to fit the video search wishes of the target users.
[0065] figure 2 It is a flow chart for determining a default search word disclosed in an embodiment of the present invention. In an illustrative example, the search term metadata includes channel popular search terms; the preference data includes channel preference data. see figure 2 , step 103 may specifically include:
[0066] Step 201: Determine the target user's preferred channel according to the channel preference data;
[0067] Users may often watch videos of a certain channel in their daily life. If the ratio of the number of times or duration of viewing this channel to the total number of times or total duration of videos watched by the user exceeds a preset threshold, then it can be determined that this channel is the user's preference channel.
[0068] Step 202: Select a search term corresponding to the preferred channel from the channel popular search terms and determine it as a default search term.
[0069] If the user prefers a certain channel, then the popular search terms of this channel are probably also the search terms that the user wants to search for. Therefore, in this example, the popular search terms of the channel under the user's preferred channel are determined as the default search term.
[0070] image 3 Another flowchart for determining a default search term disclosed in an embodiment of the present invention. In another illustrative example, the search term metadata includes frequent search terms and related search terms; the preference data includes tag preference data and search history data. see image 3 , step 103 may specifically include:
[0071] Step 301: Determine the target user's preferred tags and/or historical search terms according to the tag preference data and/or the search history data;
[0072] A user may often watch videos of a certain tag in daily life, and if the ratio of the number of times of watching videos under this tag to the total number of videos watched by the user exceeds a preset threshold, then it can be determined that this channel is the user's preferred tag.
[0073] The historical search term is a term that the user has searched before, which proves that the user is more interested in videos related to the search term.
[0074] Step 302: Determine whether each preference tag and/or historical search term is a frequent search term; if yes, proceed to step 303; if not, proceed to step 304;
[0075] Step 303: determining the preference tag and/or the historical search term as a default search term;
[0076] For the user's preferred label or historical search term, if it is a frequent search term, it proves that the user may search for the preferred label or historical search term again. In this example, in this case, it will be the user who frequently searched for the term Preference tags and historical search terms are determined as default search terms.
[0077] Step 304: If the preference tag and/or the historical search term has a related search term, determine the preference tag and/or the related search term of the historical search term as a default search term.
[0078] In the case that the preference tag or the historical search term is not a frequent search term, considering the user's preference interest, the related word of the preference tag and the historical search term is confirmed as a default search term.
[0079] In practical applications, the various values involved in the above content can be adjusted according to actual conditions, so as to control the number of generated default search words within a reasonable range.
[0080] In this embodiment, the method for generating the default search term first determines the search term metadata according to the global search history information, then determines the target user’s preference data according to the target user’s video history information, and then determines the target user’s preference data according to the search term The metadata and the preference data generate a plurality of default search terms. The default search term generation method automatically generates default search terms according to the global search history and user preference data, which is time-effective; corresponding default search terms can be generated for users with different video preferences, so that the generated default search terms can fit The needs of different users help to simplify the user's video search process and improve the user experience.
[0081] Figure 4 A flow chart of another method for generating a default search term disclosed in an embodiment of the present invention, such as Figure 4 As shown, the method may include:
[0082] Step 401: According to the global search history information, determine the channel's popular search terms, frequent search terms and related search terms with popular scores;
[0083] Wherein, the popular score may be calculated according to a first preset rule. The first preset rule may be: taking the search amount (base_imp_count) of the word with the least number of search times among the popular search words of the channel as a benchmark, the popular score of each popular search word is min(1.0,0.5+log(imp_count /base_imp_count). That is to smooth the search volume of popular search terms in each channel to the interval of [0.5,1].
[0084] Step 402: According to the target user's video history information, determine the target user's channel preference data with channel scores, tag preference data with tag scores, and search history data with search scores;
[0085] The channel score and tag score are calculated according to the second preset rule; the search score is calculated according to the fourth preset rule.
[0086] Wherein, the second preset rule may be: divide the number of views of each channel and label by the total number of views to obtain the score of the channel and label. Channels and tags with a score greater than 0.4 can be reserved to participate in the generation and determination of subsequent default search terms.
[0087]The fourth preset rule may be: count the number of searches (imp_count) and search time (imp_time) of the target user's historical search term in the past 15 days, if the target user has searched for a search term multiple times, take the latest time as search time. For each search term, calculate its search times score imp_count_score=0.74+0.13*imp_count, the more times the search term is searched, the higher the score; calculate its search freshness score imp_time_score=pow(0.85,(cur_time–imp_time)) , where cur_time is the current time, that is, the freshness score decays with a coefficient of 0.85 every day, and the longer the search term is from the current time, the lower the score; calculate the total score score=imp_count_score*imp_time_score*query_ctr, where query_ctr is the global search term click through rate.
[0088] Among them, the global click-through rate can reflect the quality of a search term, and the higher the global click-through rate, the higher the possibility of the search term being clicked by the target user.
[0089] After step 402, go to step 403 and step 406 at the same time.
[0090] Step 403: Determine the target user's preferred channel according to the channel preference data;
[0091] Step 404: Select a search word corresponding to the preferred channel from the popular search words of the channel and determine it as the default search word;
[0092] Step 405: According to the popularity score and channel score corresponding to the default search term, calculate the comprehensive score of the default search term according to the third preset rule; enter step 410;
[0093] The channel popular search term has a popular score, the channel preference data has a channel score, and the default search term takes the intersection of the two, so the determined default search term is not only a channel popular search term, but also a channel preference data, which corresponds to has a popular score and a channel score.
[0094] The third preset rule may be: multiply the popular score of the default search term by the channel score as its comprehensive score, and put it into the default search term candidate set. For example, if the popularity score corresponding to a default search term is 0.6, and the corresponding channel score is 0.5, then the comprehensive score of the default search term is 0.6*0.5=0.3.
[0095] Step 406: Determine the target user's preferred tags and/or historical search terms according to the tag preference data and/or the search history data;
[0096] Step 407: Determine whether each preference tag and/or historical search term is a frequent search term; if yes, proceed to step 408; if not, proceed to step 409;
[0097] Step 408: Determine the preference label and/or the historical search term as the default search term, and determine the preference label and/or the label score and/or search score corresponding to the historical search term as the default search term Composite score; enter step 410;
[0098] Step 409: If the preference tag and/or the historical search term have related search terms, determine the preference tag and/or the related search term of the historical search term as the default search term, and set the preference tag And/or the label score corresponding to the historical search term and/or the search score corresponding to the comprehensive score determined as the default search term; enter step 410;
[0099] Step 410: Generate a default search word list from the N default search words with the highest comprehensive scores, and provide it to the search engine.
[0100] Wherein, the N is a positive integer. After the N default search words with the highest comprehensive scores are provided to the search engine, the search engine can display the default search words in the search bar for users to view and choose.
[0101] In this embodiment, the method for generating the default search term automatically generates the default search term according to the global search history and the user's preference data, which has high timeliness; corresponding default search terms can be generated for users with different video preferences, satisfying different users It helps to simplify the user video search process and improve the user experience. Moreover, the method uses a default search term scoring mechanism, which is conducive to generating a default search term that is more in line with the user's wishes.
[0102] The method is described in detail in the above disclosed embodiments of the present invention. The method of the present invention can be realized by using various forms of devices. Therefore, the present invention also discloses a device, which will be described in detail in the following specific embodiments.
[0103] Figure 5 For a schematic structural diagram of a device for generating a default search word disclosed in an embodiment of the present invention, see Figure 5 As shown, the generating device 50 of the default search term may include:
[0104] The metadata determination module 501 is used to determine the search word metadata according to the global search history information;
[0105] According to the search history information of users on the whole network, the search word metadata can be determined. In this embodiment, the metadata may include channel popular search terms, frequent search terms and/or related search terms.
[0106] A preference data determination module 502, configured to determine the target user's preference data according to the target user's video history information;
[0107] Wherein, the preference data of the target user may be determined based on two aspects of video viewing history data and user search history data.
[0108] Using the user's search history data, it is possible to determine the search terms that the user has searched for, and determine which search terms the user is interested in.
[0109] A default word determining module 503, configured to generate a plurality of default search words according to the search word metadata and the preference data.
[0110] The determination of the default search term can combine the channel preference information and label preference information of the target user determined based on the target user’s video history information with the channel popular search terms, frequent search terms and related search terms determined based on the global search history information and search history data to determine. Based on the search situation of the entire network, the default search terms generated in combination with the personal hobbies of the target users are both public and pertinent, and it is easy to fit the video search wishes of the target users.
[0111] Image 6 A schematic diagram of the first structure of a default word determination module disclosed in an embodiment of the present invention. In an illustrative example, the search term metadata includes channel popular search terms; the preference data includes channel preference data. see Image 6 , the default word determination module 503 may include:
[0112] A first preference confirmation module 601, configured to determine a preferred channel of the target user according to the channel preference data;
[0113] Users may often watch videos of a certain channel in their daily life. If the ratio of the number of times or duration of viewing this channel to the total number of times or total duration of videos watched by the user exceeds a preset threshold, then it can be determined that this channel is the user's preference channel.
[0114] The first default word sub-module 602 is configured to select a search word corresponding to the preferred channel from the channel popular search words and determine it as a default search word.
[0115] If the user prefers a certain channel, then the popular search terms of this channel are probably also the search terms that the user wants to search for. Therefore, in this example, the popular search terms of the channel under the user's preferred channel are determined as the default search term.
[0116] Figure 7 It is a schematic structural diagram of another default word determination module disclosed in the embodiment of the present invention. In another illustrative example, the search term metadata includes frequent search terms and related search terms; the preference data includes tag preference data and search history data. see Figure 7 , the default word determination module 503 may include:
[0117] The second preference confirmation module 701 is configured to determine the target user's preferred tags and/or historical search terms according to the tag preference data and/or the search history data;
[0118] A user may often watch videos of a certain tag in daily life, and if the ratio of the number of times of watching videos under this tag to the total number of videos watched by the user exceeds a preset threshold, then it can be determined that this channel is the user's preferred tag.
[0119] The historical search term is a term that the user has searched before, which proves that the user is more interested in videos related to the search term.
[0120] A judging module 702, configured to judge whether each preference tag and/or historical search term belongs to a frequent search term;
[0121] The second default word submodule 703 is configured to determine the preference tag and/or the historical search term as a default search term when the determination result of the determination module 702 is yes; If the result is no, if the preference tag and/or the historical search term has a related search term, determine the preference tag and/or the related search term of the historical search term as a default search term.
[0122] For the user's preferred label or historical search term, if it is a frequent search term, it proves that the user may search for the preferred label or historical search term again. In this example, in this case, it will be the user who frequently searched for the term Preference tags and historical search terms are determined as default search terms.
[0123] In the case that the preference tag or the historical search term is not a frequent search term, considering the user's preference interest, the related word of the preference tag and the historical search term is confirmed as a default search term.
[0124] In practical applications, the various values involved in the above content can be adjusted according to actual conditions, so as to control the number of generated default search words within a reasonable range.
[0125] In this embodiment, the device for generating the default search term first determines the search term metadata according to the global search history information, then determines the target user’s preference data according to the target user’s video history information, and then determines the target user’s preference data according to the search term The metadata and the preference data generate a plurality of default search terms. The generating device of the default search term automatically generates the default search term according to the global search history and the user's preference data, which has high timeliness; the corresponding default search term can be generated for users with different video preferences, so that the generated default search term can fit The needs of different users help to simplify the user's video search process and improve the user experience.
[0126] Figure 8 A schematic structural diagram of another device for generating default search terms disclosed in an embodiment of the present invention, as shown in Figure 8 As shown, the generating means 80 of the default search term may include:
[0127] The metadata determination module 501 is used to determine channel popular search words, frequent search words and related search words with popular scores according to global search history information;
[0128]Wherein, the popular score may be calculated according to a first preset rule. The first preset rule may be: taking the search amount (base_imp_count) of the word with the least number of search times among the popular search words of the channel as a benchmark, the popular score of each popular search word is min(1.0,0.5+log(imp_count /base_imp_count). That is to smooth the search volume of popular search terms in each channel to the interval of [0.5,1].
[0129] The preference data determination module 502 is used to determine the target user's channel preference data with channel scores, label preference data with tag scores and search history data with search scores according to the target user's video history information;
[0130] The channel score and tag score are calculated according to the second preset rule; the search score is calculated according to the fourth preset rule.
[0131] Wherein, the second preset rule may be: divide the number of views of each channel and label by the total number of views to obtain the score of the channel and label. Channels and tags with a score greater than 0.4 can be reserved to participate in the generation and determination of subsequent default search terms.
[0132] The fourth preset rule may be: count the number of searches (imp_count) and search time (imp_time) of the target user's historical search term in the past 15 days, if the target user has searched for a search term multiple times, take the latest time as search time. For each search term, calculate its search times score imp_count_score=0.74+0.13*imp_count, the more times the search term is searched, the higher the score; calculate its search freshness score imp_time_score=pow(0.85,(cur_time–imp_time)) , where cur_time is the current time, that is, the freshness score decays with a coefficient of 0.85 every day, and the longer the search term is from the current time, the lower the score; calculate the total score score=imp_count_score*imp_time_score*query_ctr, where query_ctr is the global search term click through rate.
[0133] Among them, the global click-through rate can reflect the quality of a search term, and the higher the global click-through rate, the higher the possibility of the search term being clicked by the target user.
[0134] A first preference confirmation module 601, configured to determine a preferred channel of the target user according to the channel preference data;
[0135] The first default word sub-module 602 is used to select a search word corresponding to the preferred channel from the popular search words of the channel and determine it as the default search word;
[0136] The comprehensive score determination module 801 is used to calculate the comprehensive score of the default search term according to the third preset rule according to the popular score and channel score corresponding to the default search term;
[0137] The channel popular search term has a popular score, the channel preference data has a channel score, and the default search term takes the intersection of the two, so the determined default search term is not only a channel popular search term, but also a channel preference data, which corresponds to has a popular score and a channel score.
[0138] The third preset rule may be: multiply the popular score of the default search term by the channel score as its comprehensive score, and put it into the default search term candidate set.
[0139] The second preference confirmation module 701 is configured to determine the target user's preferred tags and/or historical search terms according to the tag preference data and/or the search history data;
[0140] A judging module 702, configured to judge whether each preference tag and/or historical search term belongs to a frequent search term;
[0141] The second default word submodule 703 is configured to determine the preference tag and/or the historical search term as a default search term when the determination result of the determination module 702 is yes; When the result is no, in the case that the preference tag and/or the historical search term has a related search term, determine the preference tag and/or the related search term of the historical search term as a default search term;
[0142] Then the comprehensive score determination module 801 is further configured to, when the judgment result of the judgment module is yes, determine the tag score and/or search score corresponding to the preference tag and/or the historical search term as the corresponding The comprehensive score of the default search term; when the judgment result of the judgment module is no, in the case that the preference tag and/or the history search term have related search words, the preference tag and/or the history search term The tag score and/or search score corresponding to the word is determined as the composite score of the default search word;
[0143] The list generating module 802 is configured to generate a default search word list from the N default search words with the highest comprehensive scores, and provide it to the search engine.
[0144] Wherein, the N is a positive integer. After the N default search words with the highest comprehensive scores are provided to the search engine, the search engine can display the default search words in the search bar for users to view and choose.
[0145] In this embodiment, the device for generating default search words automatically generates default search words according to the global search history and user preference data, which is time-effective; corresponding default search words can be generated for users with different video preferences to meet the needs of different users. It helps to simplify the user video search process and improve the user experience. Moreover, the default search term scoring mechanism is cited in the device, which is conducive to generating a default search term that is more in line with the user's wishes.
[0146] Each embodiment in this specification is described in a progressive manner, each embodiment focuses on the difference from other embodiments, and the same and similar parts of each embodiment can be referred to each other. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and for the related information, please refer to the description of the method part.
[0147] It should also be noted that in this article, relational terms such as first and second etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that these entities or operations Any such actual relationship or order exists between. Furthermore, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article, or apparatus comprising a set of elements includes not only those elements, but also includes elements not expressly listed. other elements of or also include elements inherent in such a process, method, article, or device. Without further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising said element.
[0148] The steps of the methods or algorithms described in connection with the embodiments disclosed herein may be directly implemented by hardware, software modules executed by a processor, or a combination of both. Software modules can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other Any other known storage medium.
[0149] The above description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
