WO2018000569A1 - 话题订阅方法、装置和存储介质 - Google Patents
话题订阅方法、装置和存储介质 Download PDFInfo
- Publication number
- WO2018000569A1 WO2018000569A1 PCT/CN2016/097477 CN2016097477W WO2018000569A1 WO 2018000569 A1 WO2018000569 A1 WO 2018000569A1 CN 2016097477 W CN2016097477 W CN 2016097477W WO 2018000569 A1 WO2018000569 A1 WO 2018000569A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- resource
- topic
- persistent
- user
- persistent topic
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/17—Details of further file system functions
- G06F16/174—Redundancy elimination performed by the file system
- G06F16/1748—De-duplication implemented within the file system, e.g. based on file segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/31—Indexing; Data structures therefor; Storage structures
- G06F16/313—Selection or weighting of terms for indexing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3347—Query execution using vector based model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Definitions
- the embodiments of the present invention relate to the field of Internet technologies, and in particular, to a topic subscription method, apparatus, and storage medium.
- the topic subscription methods in the prior art are roughly divided into two types, one is based on resources, and the other is based on demand subscription, which are respectively introduced as follows:
- the resource-based recommendation method uses the headline of today as an example to guess the user's interest by recording the user's daily click resources, and implicitly recommend related resources. For example, if a user frequently clicks on an NBA-related article, the user is Continued use of the "Today's headlines” process will reveal a significant increase in other NBA-related articles.
- the resources of “Today's headlines” are mainly provided by the cooperative media. Technically, through certain screenings, the resources that meet the requirements are filtered out, and the similarity between the resources is used to make implicit recommendations. "Today's headlines" also support the function of subscription. First, users mainly subscribe to topics through entity words.
- the granularity of demand in this way is too coarse, and the probability that the articles in the subscription will hit the real needs of users will be lower;
- the media number is used to subscribe to the topic. This method relies too much on the media number and cannot consistently hit the needs of a certain user.
- the recommendation method based on demand subscription is “immediately”.
- the “immediate” homepage is a recommendation for some topic topics. Users can search for categories or directly search for topics of interest. After the user subscribes to certain demand topics, "immediately” pushes the user to remind them that the subscription has new resources. The user can also click on the "message" to view the latest resources in the subscription requirements, as shown in Figure 2.
- a main object of embodiments of the present invention is to provide a topic subscription method, apparatus, and storage medium to increase the probability of hitting a topic of interest to a user.
- an embodiment of the present invention provides a topic subscription method, where the method includes:
- an embodiment of the present invention further provides a topic subscription device, where the device includes:
- a topic matching module configured to match the search keyword to the persistent topic according to the user's search keyword, combined with the user's historical behavior and/or subscription record;
- a topic display module configured to return the persistent topic to a client display for the user to subscribe to;
- the resource recommendation module is configured to save the persistent topic of the user subscription, and recommend the updated matching resource to the user when the matching resource corresponding to the persistent topic subscribed by the user is updated.
- an embodiment of the present invention further provides a storage medium including computer executable instructions for executing a topic subscription method when executed by a computer processor, the method comprising:
- the topic subscription method, apparatus, and storage medium provided in the embodiments of the present invention match the persistent topic by combining the user's search keyword with the user's historical behavior and/or subscription record, and the persistent message is
- the problem returns to the client display for the user to subscribe, and when the matching resource corresponding to the persistent topic subscribed by the user is updated, the updated matching resource is recommended to the user, because the user can adjust the recommendation according to the real-time behavior of the user in combination with the historical behavior.
- the strategy increases the probability of hitting a topic of interest to the user.
- FIG. 1 is an exemplary diagram of an instant subscription page in the prior art
- FIG. 2 is a diagram showing an example of an instant push message in the prior art
- FIG. 3 is a flowchart of a topic subscription method according to Embodiment 1 of the present invention.
- 4a-4c are diagrams showing an example of a subscription page in a topic subscription method according to an embodiment of the present invention.
- FIG. 5 is a flowchart of a topic subscription method according to Embodiment 2 of the present invention.
- FIG. 6 is a diagram showing an example of a matching resource and a corresponding persistent topic recommended by a user according to a search keyword of a user in a topic subscription method according to an embodiment of the present invention
- FIG. 7 is a diagram showing an example of updated matching resources in a topic subscription method according to an embodiment of the present invention.
- FIG. 8 is a flowchart of a topic subscription method according to Embodiment 3 of the present invention.
- FIG. 9 is a flowchart of establishing an association between a persistent topic and a resource in a topic subscription method according to Embodiment 4 of the present invention.
- FIG. 10 is a schematic structural diagram of a topic subscription apparatus according to Embodiment 5 of the present invention.
- FIG. 11 is a schematic structural diagram of a server according to Embodiment 6 of the present invention.
- Embodiment 3 is a flowchart of a topic subscription method according to Embodiment 1 of the present invention.
- the present embodiment is applicable to a situation in which a topic is matched according to a historical behavior of a user and is subscribed by a user.
- the method may be performed by a topic subscription device.
- the topic subscription device can be configured in the server, and the method specifically includes the following:
- the search keyword can be input to retrieve related topics and corresponding resources.
- the search keyword is combined with the user's historical behavior (such as the user's click resource and the user's search history, etc., that is, the user's session history) And/or a subscription record (a topic of the user's subscription) that matches the search term to a persistent topic of interest to the user that is a topic of continuing need for the user.
- the persistent topic of matching includes at least one persistent topic.
- a persistent topic related to the search keyword input by the user as a related recommendation.
- search keyword There are two cases related to the search keyword, including subject-related or keyword-related. For example, when a user searches for "Luhan Girlfriend", the user firstly matches the persistent topic of the user's needs, and then recommends the subject's "Luhan Run Man 4" and "Wang Zulan Height” (with the Luhan team participating in the running man, Other topics such as "Wu Yifan girlfriend” and "Zheng Zheng girlfriend” related to keywords.
- These related recommendations will help users discover the needs around the main demand, so that more demand and its associated quality resources will be seen by users, inspiring the user's other potential needs.
- the persistent topic includes at least one user query directly or indirectly related to the set time interval, and is a topic stored in the persistent topic search library.
- the association between the resource and the persistent topic in the persistent topic search database is established, and when the resource is determined to be a matching resource corresponding to the persistent topic subscribed by the user, the updated matching resource may be in the form of a message. Push to the client.
- FIG. 4a-4c are diagrams showing an example of a subscription page in a topic subscription method according to an embodiment of the present invention.
- each subscription is composed of a persistent topic and a title of the corresponding latest resource
- the subscription may also include a picture
- the subscription page also supports the user to search for what he wants.
- Subscription content As shown in Figure 4b, when users subscribe to some of these persistent topics, you can click "My" to view the persistent topics that have been subscribed to.
- Figure 4c the user can also click on "message” to view the latest resources of the persistent topic that has been subscribed in a waterfall flow.
- the technical solution of this embodiment matches the persistent topic by combining the user's search keyword with the user's historical behavior and/or subscription record, and returns the persistent topic to the client display for the user to subscribe, and subscribes to the user.
- the matching resource corresponding to the persistent topic is updated, the updated matching resource is recommended to the user. Since the user can adjust the recommended policy according to the real-time behavior of the user in combination with the historical behavior, the probability of hitting the topic of interest to the user is increased, and the user is improved. Discover more experiences on ongoing topics.
- the topic subscription method further preferably includes:
- a persistent topic recommended to the user is determined and recommended to the user according to the historical behavior of the user and the subscription behavior of the persistent topic.
- the matching resource corresponding to the persistent topic is returned to the client for display. For example, when the current user just opens the application, the corresponding persistent topic is recommended to the current user according to the current user's historical behavior and the number of subscribers of the persistent topic. The purpose of recommending to the user a topic of interest to the user and high popularity is achieved.
- matching the persistent topic for the search keyword preferably includes:
- the persistent topics are sorted according to the click record of the persistent topic and the historical behavior of the user.
- the user's historical behavior and/or subscription record is combined to determine a persistent topic that matches the search keyword, and then according to the determined persistent topic click record and the user's historical behavior, The continuous topic is sorted, and the persistent topic with the number of click records and the user's historical behavior is ranked first, so that the user can recommend the persistent topic with high popularity and interest to the user according to the sorting result, and further increase The probability of hitting a topic of interest to the user.
- FIG. 5 is a flowchart of a topic subscription method according to Embodiment 2 of the present invention.
- the embodiment is optimized based on the foregoing embodiment, and further adds “in the case of receiving a persistent topic of user clicks, in advance
- the established persistent topic retrieval library queries the corresponding matching resource and returns the client display, and the method specifically includes the following:
- the persistent topic retrieval library includes an association relationship between a persistent topic and a matching resource.
- the persistent topic in the persistent topic retrieval library is a user query that meets a preset condition based on the retrieval logs of all users.
- the client detects the click of the user in real time.
- the client sends the persistent topic clicked by the client to the server, and the server queries the matching topic corresponding to the persistent topic in the pre-established persistent topic search library. Return the queried matching resource to the client display.
- the matching resource is returned to the client display.
- the matching resource corresponding to the persistent topic is queried in the pre-established persistent topic search library, according to the click record and the resource attribute (such as the value score, etc.) of the matched matching resource. And the user's historical behavior, sorting the matching resources in descending order, and returning the matching resources to the client display according to the sorting result, so that the user can first obtain the resources with high heat, good quality, and user interest, The probability of hitting resources of interest to the user is further increased.
- FIG. 6 is a diagram showing an example of a matching resource and a corresponding persistent topic recommended by a user according to a search keyword of a user in a topic subscription method according to an embodiment of the present invention.
- FIG. 6 when the user inputs “star gossip” First, recommend some of the more popular star satisfaction resources 610, and then recommend the more popular continuous topic 620 with persistent demand; when the continuous topic "star gossip” is updated, you can pass the middle shown in Figure 7. Click to browse related resources.
- FIG. 7 is a diagram showing an example of updated matching resources in a topic subscription method according to an embodiment of the present invention.
- the corresponding matching resource when receiving the persistent topic clicked by the user, the corresponding matching resource is queried in the pre-established persistent topic search library, and returned to the client for display.
- the user views the matching resources corresponding to the persistent topic.
- FIG. 8 is a flowchart of a topic subscription method according to Embodiment 3 of the present invention.
- the embodiment is optimized on the basis of the foregoing embodiment, and the operation of establishing a persistent topic search library under online is added, and the method specifically includes as follows:
- the preset condition preferably includes: continuously searching for a second preset number of days beyond a preset number of people within a first preset number of days, and/or including a preset keyword.
- the user query that meets the preset condition is mined as a persistent topic. For example, in 2 weeks, 10 users searched for "Huang Huaweing angelababy" for more than 3 consecutive days, then "Huang Huaweing angelababy” was a continuous topic; user queries with preset keywords, such as "latest TV series” with pre- With the keyword “latest”, the "latest TV series” is a continuous topic.
- the excavated continuous topic is literally de-emphasized as a result of all continuing topics. It is also possible to combine the latest search logs of all users and routinely explore the latest ongoing topics.
- the resource pool is established through resource mining, and the source of the resource mining includes a pre-configured website and a click resource of the persistent topic.
- Pre-configured websites are some of the more high-quality topic websites, such as “Today's headlines”, “Knowing”, etc., to obtain resources from these topic websites, and the click resources of the persistent topics mentioned in this website, combining these two sources. Resources, build a resource library.
- establishing a resource library preferably includes:
- Extracting resource attributes of the resource including a title, a summary, a map, and a value score;
- the resource and corresponding resource attributes are stored in the resource library.
- the title in the resource attribute can be obtained by directly capturing the title of the resource; the abstract is obtained by extracting the key sentence in the title and the body; determining the resource body according to the original picture of the resource, and then performing the preset size on the main picture
- the cropping is matched with the map; the value score of the resource is mainly determined by the amount and depth of the information of the resource content, and the source of the resource. In general, the value of the content with novel and deep content and high image content is higher.
- the association of the persistent topic with the resources in the repository is established based on the word distribution of the persistent topic and the word distribution of the resources in the repository.
- the persistent topic with potential potential needs of the user is automatically mined, and valuable value is mined.
- the resource establishes the association between the persistent topic and the resource, and stores the persistent topic retrieval library, so as to facilitate subsequent recommendation to the user for the persistent topic and the corresponding resource that the user is interested in, and to mine with respect to the prior art through the topic mining.
- the way to organize topics can provide more sustainable topics that can be subscribed, and have more relevant resources to support and satisfy user subscriptions.
- FIG. 9 is a persistent topic and a resource in a topic subscription method according to Embodiment 4 of the present invention.
- the flow chart of the association of the source, the embodiment is further optimized on the basis of the third embodiment, "establishing the association between the persistent topic and the resource in the resource library, and depositing the persistent topic retrieval library".
- the “establishing the association between the persistent topic and the resource in the resource library and storing the persistent topic search database” in the embodiment includes the following:
- the word distribution of the resource is obtained by counting the title of the resource and the word frequency in the body.
- the similarity between the persistent topics is calculated according to the word distribution and the word-embedding feature of the persistent topic, and the persistent topic is deduplicated and clustered according to the similarity.
- deduplication and clustering processing on the persistent topic includes:
- the persistent topic whose similarity is greater than the second preset threshold and less than or equal to the first preset threshold is classified into one category.
- the first preset threshold is greater than the second preset threshold.
- Clustering refers to classifying persistent topics with similar semantic meanings, such as "Wang Yuan Photo” and "Wang Yuan Street Shooting”. The deduplication and clustering of persistent topics can be improved by the similarity between persistent topics, which can improve the accuracy of processing.
- the established relationship between the persistent topic and the resource is established, and the deposit into the persistent topic retrieval library preferably includes:
- the candidate persistent topic with the similarity greater than the preset threshold is stored as a persistent topic associated with the resource in the persistent topic search library.
- Inverted index also often referred to as reverse index, placed file or reverse file
- reverse index is an index method used to store a word in a document or a group under full-text search.
- the mapping of storage locations in the document By inverting the index, you can quickly get a list of documents containing the word based on the word.
- the candidate persistent topic includes at least one item, and the similarity between the word with more trips in the resource body and the word distribution of the candidate persistent topic is calculated.
- the candidate persistent topic whose similarity is greater than the preset threshold is used as a persistent topic associated with the resource, and all the persistent topics and resource associations are determined and stored in the persistent topic search library.
- the resource titled “The Mask in the Descendants of the Sun” has repeatedly mentioned “mask” and “hydration” in the text.
- the similarity between these words and the persistent topic "descendants of the sun” is very low, so it is determined.
- the resource does not match the “descendants of the sun”;
- the resource titled “Reviewing the Classical Picture of the Sun's Descendants” has repeatedly mentioned “the descendants of the sun”, “Song Zhongji”, and “Song Huiqiao” in the text.
- the relevance of the topic "The descendants of the sun” is relatively high, so it is determined that the resource can match the "descendants of the sun”.
- S834 De-duplicate the resource associated with the processed persistent topic according to the word distribution of the resource associated with the persistent topic.
- the deduplication cost of resources directly on the full amount of resources will be relatively high. Therefore, based on the matching results of persistent topics and resources, the resources associated with each persistent topic are deduplicated.
- the word distribution of the resource title between the two resources and the similarity of the word distribution of the resource body are mainly considered to perform deduplication.
- the deduplication of the resource associated with the processed persistent topic preferably includes:
- resources A, B, and C have been reserved, and the resource D is streamed. It is found that D is similar to A, and the value score of D is higher than the value score of A. Then resource D is replaced with resource A, that is, resource D is reserved. B, C.
- the comparison speed can be improved by means of streaming comparison.
- the newly added resource can be connected to the continuation in time.
- the resources associated with the persistent topic are deduplicated, and the speed of deduplication can be improved.
- the topic subscription device in this embodiment includes: a topic matching module 1010, a topic display module 1020, and a resource recommendation module 1030. .
- the topic matching module 1010 is configured to match the search keyword to the persistent topic according to the user's search keyword, combined with the user's historical behavior and/or subscription record;
- the topic recommendation module 1020 is configured to return the persistent topic to the client for display by the user;
- the resource recommendation module 1030 is configured to save the persistent topic subscribed by the user, and recommend the updated matching resource to the user when the matching resource corresponding to the persistent topic subscribed by the user is updated.
- the method further includes:
- a topic recommendation module for subscribing behavior according to the historical behavior of the user and the persistent topic, A persistent topic recommended to the user is determined and recommended to the user.
- the method further includes:
- a matching resource display module configured to query a corresponding matching resource in a pre-established persistent topic retrieval library when receiving a persistent topic clicked by the user, and return a client display, where the persistent topic retrieval library includes The relationship between persistent topics and matching resources.
- the matching resource display module includes:
- a matching resource query unit configured to query a corresponding matching resource in a pre-established persistent topic search library when receiving a persistent topic clicked by the user;
- a matching resource sorting unit configured to sort the matching resources according to the click record, the resource attribute, and the historical behavior of the user of the matching resource
- the matching resource recommendation unit is configured to return the matching resource to the client display according to the sorting result.
- the topic matching module includes:
- the topic determining unit determines a persistent topic of the search keyword according to the user's search keyword, combined with the user's historical behavior and/or subscription record;
- the topic sorting unit is configured to sort the persistent topic according to the click record of the persistent topic and the historical behavior of the user.
- the method further includes:
- a topic mining module configured to mine a user that meets a preset condition according to a user search log before matching the persistent keyword to the search keyword according to the user's search keyword and the user's historical behavior and/or subscription record Query as a continuous topic;
- a resource mining module configured to establish a resource pool according to a pre-configured website and a click resource of the persistent topic
- the topic resource association module is configured to establish an association between the persistent topic and resources in the resource library, and store the persistent topic search library.
- the preset condition preferably includes: continuously searching for a second preset number of days beyond a preset number of people within a first preset number of days, and/or including a preset keyword.
- the resource mining module includes:
- An original resource obtaining unit configured to acquire resources and the persistent words in the pre-configured website Click on the resource as the original resource
- a resource literal deduplication unit configured to decrement the original resource with the same literal content according to the title and the body of the original resource, to obtain a resource
- a resource attribute extraction unit configured to extract a resource attribute of the resource, where the resource attribute includes a title, a summary, a map, and a value score;
- a resource library establishing unit configured to store the resource and the corresponding resource attribute into the resource library.
- the topic resource association module includes:
- a word distribution extracting unit configured to respectively extract the persistent topic and a word distribution of the resource, wherein the statistical corpus of the persistent topic includes a click resource of the persistent topic and a title in the resource library includes the persistent Resources for sexual topics;
- a topic processing unit configured to perform deduplication and clustering processing on the persistent topic according to the word distribution and the word vector feature of the persistent topic
- the topic resource association unit is configured to establish a relationship between the processed persistent topic and the resource, and store the persistent topic search database;
- the resource word distribution deduplication unit is configured to de-duplicate the resources associated with the processed persistent topic according to the word distribution of the resources associated with the persistent topic.
- the topic processing unit is specifically configured to:
- the persistent topic whose similarity is greater than the second preset threshold and less than or equal to the first preset threshold is classified into one category.
- the topic resource association unit is specifically configured to:
- the candidate persistent topic with the similarity greater than the preset threshold is stored as a persistent topic associated with the resource in the persistent topic search library.
- the resource word deduplication unit is specifically used to:
- the above product can perform the method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
- the sixth embodiment of the present invention provides a server, including the topic subscription device provided by any embodiment of the present invention.
- an embodiment of the present invention provides a server, where the server includes a processor 1110, a memory 1120, an input device 1130, and an output device 1140.
- the number of processors 1110 in the server may be one or more.
- a processor 1110 is taken as an example; the processor 1110, the memory 1120, the input device 1130, and the output device 1140 in the server may be connected by a bus or other means, and the bus connection is taken as an example in FIG.
- the memory 1120 is used as a computer readable storage medium, and can be used to store a software program, a computer executable program, and a module, such as a program instruction/module corresponding to the topic subscription method in the embodiment of the present invention (for example, topic matching in a topic subscription device) Module 1010, topic display module 1020, and resource recommendation module 1030).
- the processor 1110 executes various functional applications and data processing of the server by executing software programs, instructions, and modules stored in the memory 1120, that is, implementing the above-described topic subscription method.
- the memory 1120 may mainly include a storage program area and an storage data area, wherein the storage program area may store an operating system, an application required for at least one function; the storage data area may store data created according to usage of the server, and the like.
- memory 1120 can include high speed random access memory, and can also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device.
- memory 1120 can further include memory remotely located relative to processor 1110, which can be connected to the server over a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
- Input device 1130 can be used to receive input numeric or character information and to generate key signal inputs related to user settings and function control of the server.
- the output device 1140 may include a display device such as a display screen.
- Embodiments of the present invention also provide a storage medium including computer executable instructions for executing a topic subscription method when executed by a computer processor, the method comprising:
- the present invention can be implemented by software and necessary general hardware, and can also be implemented by hardware, but in many cases, the former is a better implementation. .
- the technical solution of the present invention which is essential or contributes to the prior art, may be embodied in the form of a software product, which may be stored in a computer readable storage medium, such as a floppy disk of a computer. , Read-Only Memory (ROM), Random Access Memory (RAM), Flash (FLASH), hard disk or optical disk, etc., including a number of instructions to make a computer device (can be a personal computer)
- the server, or network device, etc. performs the methods described in various embodiments of the present invention.
- each unit and module included in the foregoing is only divided according to functional logic, but is not limited to the above-mentioned division, as long as the corresponding functions can be implemented;
- the specific names of the functional units are also for convenience of distinguishing from each other and are not intended to limit the scope of the present invention.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
一种话题订阅方法、装置和存储介质。该方法包括:根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题(S310);将所述持续性话题返回客户端显示,以供所述用户进行订阅(S320);保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户(S330)。由于能够根据用户实时的行为结合历史行为及时的调整推荐的策略,增加了命中用户感兴趣的话题的概率。
Description
本专利申请要求于2016年6月27日提交的、申请号为201610481888.4、申请人为北京百度网讯科技有限公司、发明名称为“话题订阅方法和装置”的中国专利申请的优先权,该申请的全文以引用的方式并入本申请中。
本发明实施例涉及互联网技术领域,尤其涉及一种话题订阅方法、装置和存储介质。
现有技术中的话题订阅方式大致分成两种,一种是以资源为主,一种是以需求订阅为主,分别介绍如下:
以资源为主的推荐方式以今日头条为例,通过记录用户日常的点击资源来猜测用户的兴趣,隐式的推荐相关的资源,例如,某个用户经常点击NBA相关的文章,则该用户在持续使用“今日头条”的过程中,会发现其它NBA相关的文章明显的增多。“今日头条”的资源主要由合作媒体来提供,技术上通过一定的筛选过滤出满足要求的资源,通过计算资源之间的相似度来进行隐式推荐。“今日头条”也支持订阅的功能,一是用户主要通过实体词来订阅话题,这种方式的需求粒度太粗,该订阅中的文章命中用户真正需求的概率就会比较低;二是用户通过媒体号来订阅话题,这种方式太依赖该媒体号,不能持续的命中某个用户的需求。
以需求订阅为主的推荐方式以“即刻”为例,如图1所示,在“即刻”首页是一些需求话题订阅的推荐,用户可以分类查找,或者直接搜索自己感兴趣的话题。当用户订阅某些需求话题之后,“即刻”通过推送来提醒用户这些订阅有新的资源,用户也可以通过点击“消息”来查看订阅需求里的最新的资源,如图2所示。
综上所述,以资源为主和以需求订阅为主的方式都存在着相同的缺陷,即
没有考虑用户自身的属性或者行为,不能很好的表示用户真正的兴趣点。
发明内容
本发明实施例的主要目的在于,提供一种话题订阅方法、装置和存储介质,以增加命中用户感兴趣的话题的概率。
第一方面,本发明实施例提供了一种话题订阅方法,所述方法包括:
根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;
将所述持续性话题返回客户端显示,以供所述用户进行订阅;
保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
第二方面,本发明实施例还提供了一种话题订阅装置,所述装置包括:
话题匹配模块,用于根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;
话题显示模块,用于将所述持续性话题返回客户端显示,以供所述用户进行订阅;
资源推荐模块,用于保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
第三方面,本发明实施例还提供了一种包含计算机可执行指令的存储介质,所述计算机可执行指令在由计算机处理器执行时用于执行一种话题订阅方法,该方法包括:
根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;
将所述持续性话题返回客户端显示,以供所述用户进行订阅;
保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
本发明实施例中提供的话题订阅方法、装置和存储介质,通过将用户的检索关键词结合用户的历史行为和/或订阅记录来匹配持续性话题,将该持续性话
题返回客户端显示以供用户进行订阅,并在用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户,由于能够根据用户实时的行为结合历史行为及时的调整推荐的策略,增加了命中用户感兴趣的话题的概率。
为了更清楚地说明本发明实施例中的技术方案,下面将对实施例描述中所需使用的附图作简单地介绍,当然,以下描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以对这些附图进行修改和替换。
图1是现有技术中的即刻的订阅页面的示例图;
图2是现有技术中的即刻推送的消息的示例图;
图3是本发明实施例一提供的一种话题订阅方法的流程图;
图4a-4c是本发明实施例提供的话题订阅方法中的订阅页面的示例图;
图5是本发明实施例二提供的一种话题订阅方法的流程图;
图6是本发明实施例提供的话题订阅方法中的根据用户的检索关键词为用户推荐的匹配资源和相应的持续性话题的示例图;
图7是本发明实施例提供的话题订阅方法中的更新的匹配资源的示例图;
图8是本发明实施例三提供的一种话题订阅方法的流程图;
图9是本发明实施例四提供的一种话题订阅方法中的建立持续性话题与资源的关联的流程图;
图10是本发明实施例五提供的一种话题订阅装置的结构示意图;
图11是本发明实施例六提供的一种服务器的结构示意图。
下面将结合附图对本发明实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本发明的一部分实施例,而不是全部的实施例,是为了阐述本发明的原理,而不是要将本发明限制于这些具体的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有
其他实施例,都属于本发明保护的范围。
实施例一
图3是本发明实施例一提供的一种话题订阅方法的流程图,本实施例可适用于根据用户的历史行为匹配话题并供用户订阅的情况,该方法可以由话题订阅装置来执行,该话题订阅装置可以做配置在服务器中,该方法具体包括如下:
S310,根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题。
当用户使用话题检索的应用程序(如百度度秘明星八卦服务)时,可以输入检索关键词来检索相关的话题及对应的资源。在接收到用户输入的检索关键词时,为了更好的命中用户的兴趣点,将所述检索关键词结合用户的历史行为(如用户的点击资源和用户的检索历史等,即用户的会话历史)和/或订阅记录(用户的订阅的话题),为所述检索关键词匹配用户感兴趣的持续性话题,该持续性话题对用户来说是有持续性需求的话题。其中,匹配的持续性话题包括至少一项持续性话题。
在匹配持续性话题时,还可以匹配与用户输入的检索关键词相关的持续性话题,作为相关的推荐。与检索关键词相关包括主体相关或关键词相关两种情况。例如,用户搜索“鹿晗女朋友”时,首先精准匹配出用户需求的持续性话题,然后推荐主体相关的“鹿晗跑男4的”、“王祖蓝身高”(与鹿晗共同参与跑男,好朋友关系)等话题,和关键词相关的“吴亦凡女朋友”、“郑恺女朋友”等话题。这些相关的推荐会帮助用户发现主需求周边的需求,进而让更多需求和其关联的优质资源被用户看见,激发出用户潜在的其它需求。
其中,所述持续性话题包括设定时间区间内,直接或者间接相关的至少一项用户查询,是存储在持续性话题检索库中的话题。
S320,将所述持续性话题返回客户端显示,以供所述用户进行订阅。
将所述持续性话题返回给用户的客户端,在客户端的显示界面进行显示,如果用户点击其中的一个持续性话题,则在预先建立的持续性话题检索库中查询与该持续性话题对应的匹配资源,并返回客户端显示,以供用户浏览该持续性话题对应的匹配资源。
S330,保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
接收所述用户订阅的持续性话题,并保存在所述用户的账户下。当有资源更新时,建立该资源与持续性话题检索库中的持续性话题的关联,确定该资源是用户订阅的持续性话题对应的匹配资源时,可以以消息的形式将该更新的匹配资源推送到客户端。
图4a-4c是本发明实施例提供的话题订阅方法中的订阅页面的示例图。如图4a所示,在订阅页面的热门订阅中,每个订阅项由持续性话题和对应的最新的资源的标题构成,所述订阅项还可以包括图片,订阅页面还支持用户搜索自己想要的订阅内容。如图4b所示,当用户订阅其中一些持续性话题时,可以点开“我的”查看已经订阅的持续性话题。如图4c所示,用户还可以点开“消息”以瀑布流的方式查看已经订阅的持续性话题的最新资源。
本实施例的技术方案,通过将用户的检索关键词结合用户的历史行为和/或订阅记录来匹配持续性话题,将该持续性话题返回客户端显示以供用户进行订阅,并在用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户,由于能够根据用户实时的行为结合历史行为及时的调整推荐的策略,增加了命中用户感兴趣的话题概率,提升了用户发现更多的持续性话题的体验。
在上述技术方案的基础上,所述话题订阅方法还优选包括:
根据所述用户的历史行为及持续性话题的订阅行为,确定向所述用户推荐的持续性话题并推荐给所述用户。
根据当前用户的历史行为以及所有用户中持续性话题的订阅人数,确定向当前用户推荐的持续性话题,并将该持续性话题推荐给当前用户,在检测到用户点击该持续性话题时,将该持续性话题对应的匹配资源返回给客户端进行显示。例如,当前用户刚刚打开应用程序时,根据当前用户的历史行为以及持续性话题的订阅人数,向当前用户推荐相应的持续性话题。实现了向用户推荐用户感兴趣且热度高的话题的目的。
在上述技术方案的基础上,根据用户的检索关键词,结合用户的历史行为
和/或订阅记录,为所述检索关键词匹配持续性话题优选包括:
根据用户的检索关键词,结合用户的历史行为和/或订阅记录,确定所述检索关键词的持续性话题;
根据所述持续性话题的点击记录和用户的历史行为,对所述持续性话题进行排序。
首先根据用户的检索关键词结合用户的历史行为和/或订阅记录,确定与所述检索关键词匹配的持续性话题,再根据确定的持续性话题的点击记录和所述用户的历史行为,对所述持续性话题进行排序,将点击记录次数多且和用户的历史行为最接近的持续性话题排在前面,从而可以按照排序结果为用户推荐热度高且用户感兴趣的持续性话题,进一步增加了命中用户感兴趣的话题的概率。
实施例二
图5是本发明实施例二提供的一种话题订阅方法的流程图,本实施例在上述实施例的基础上进行了优化,进一步增加了“在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示”,该方法具体包括如下:
S510,根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题。
S520,将所述持续性话题返回客户端显示,以供所述用户进行订阅。
S530,保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
S540,在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示。
其中,所述持续性话题检索库包括持续性话题和匹配资源的关联关系。所述持续性话题检索库中的持续性话题是根据所有用户的检索日志,挖掘出的满足预设条件的用户查询(query)。
在客户端中显示所述持续性话题(通过检索关键词匹配的持续性话题返回客户端显示或者显示用户订阅的持续性话题)时,客户端实时检测用户点击的
持续性话题,在检测到用户点击的持续性话题时,客户端将用户点击的持续性话题发送给服务器,服务器在预先建立的持续性话题检索库中查询与该持续性话题对应的匹配资源,将查询到的匹配资源返回客户端显示。
其中,在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示优选包括:
在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源;
根据所述匹配资源的点击记录、资源属性和用户的历史行为,对所述匹配资源进行排序;
按照排序结果,将所述匹配资源返回客户端显示。
在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询与该持续性话题对应的匹配资源,根据查询到的匹配资源各自的点击记录、资源属性(如价值得分等)和用户的历史行为,对所述匹配资源进行降序排序,并按照排序结果,将所述匹配资源返回客户端显示,从而用户可以最先获取到热度高、质量好且用户感兴趣的资源,进一步增加了命中用户感兴趣的资源的概率。
本发明实施例提供的话题订阅方法,还可以根据用户的检索关键词直接将最接近的匹配资源和相关的持续性话题推荐给用户。图6是本发明实施例提供的话题订阅方法中的根据用户的检索关键词为用户推荐的匹配资源和相应的持续性话题的示例图,如图6所示,当用户输入“明星八卦”时,首先推荐一些比较热门明星的满足资源610,然后推荐出比较热门的具有持续性需求的明星的持续性话题620;当持续性话题“明星八卦”有更新时,可以通过图7所示的中间页来浏览相关资源。图7是本发明实施例提供的话题订阅方法中的更新的匹配资源的示例图。
本实施例的技术方案,在上述实施例的基础上,通过在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示,供用户查看持续性话题对应的匹配资源。
实施例三
图8是本发明实施例三提供的一种话题订阅方法的流程图,本实施例在上述实施例的基础上进行了优化,增加了在线下建立持续性话题检索库的操作,该方法具体包括如下:
S810,根据用户检索日志,挖掘出满足预设条件的用户查询作为持续性话题。
其中,所述预设条件优选包括:在第一预设天数内超过预设人数连续搜索第二预设天数,和/或包括预设关键词。
根据用户检索日志(如百度全网的用户检索日志),挖掘出满足预设条件的用户查询作为持续性话题。例如,在2周内有10个用户连续超过3天都搜索过“黄晓明angelababy”,则“黄晓明angelababy”为持续性话题;带有预设关键词的用户查询,如“最新电视剧”带有预设关键词“最新”,则“最新电视剧”为持续性话题。将挖掘出的持续性话题进行字面去重后,作为所有的持续性话题的结果。还可以结合所有用户最新的检索日志,例行挖掘最新的持续性话题。
S820,根据预先配置网站和所述持续性话题的点击资源,建立资源库。
通过资源挖掘来建立资源库,资源挖掘的来源包括预先配置网站和所述持续性话题的点击资源。预先配置网站为一些比较优质的话题网站,如“今日头条”、“知乎”等,从这些话题网站来获取资源,以及本网站中所述持续性话题的点击资源,综合这两种来源的资源,建立资源库。
其中,根据预先配置网站和所述持续性话题的点击资源,建立资源库优选包括:
获取所述预先配置网站中的资源和所述持续性话题的点击资源,作为原始资源;
根据所述原始资源的标题和正文,对字面内容一致的原始资源进行去重,得到资源;
提取所述资源的资源属性,所述资源属性包括标题、摘要、配图和价值得分;
将所述资源和对应的资源属性存入资源库。
其中,所述资源属性中的标题可以直接抓取资源的标题来获得;摘要通过提取标题和正文中的关键句获得;根据资源的原始图片来确定资源主体,然后通过对主图进行预设尺寸的裁剪得到配图;资源的价值得分主要依赖资源内容的信息量和深度,以及资源的来源来确定,一般而言,内容新颖有深度且图文并茂的资源的价值得分越高。通过对资源进行字面去重,可以防止资源明显的重复,通过将资源和对应的资源属性存入资源库,有利于后续跟进资源属性来建立资源与持续性话题的关联,提高关联速度。
S830,建立所述持续性话题和所述资源库中的资源的关联,存入持续性话题检索库。
根据持续性话题的词分布和所述资源库中的资源的词分布,来建立所述持续性话题和所述资源库中的资源的关联。通过计算持续性话题的词分布和所述资源库中的资源的词分布的相似度,在相似度大于预设阈值时,确定持续性话题与资源是关联的,将所述持续性话题和所述资源库中的资源的关联关系,存入持续性话题检索库,以向用户推荐持续性话题和与该持续性话题相关的资源。
S840,根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题。
S850,将所述持续性话题返回客户端显示,以供所述用户进行订阅。
S860,保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
本实施例的技术方案,在上述技术方案的基础上,通过以用户检索日志和预先配置网站为基础,自动的挖掘出有用户潜在可能的持续性需求的持续性话题,并挖掘出有价值的资源,建立所述持续性话题与资源的关联,存入持续性话题检索库,便于后续向用户推荐用户感兴趣的持续性话题和对应的资源,而且通过话题挖掘相对于现有技术中的人工整理话题的方式,可以提供更多的可订阅的持续性话题,且具备更多的相关资源来支持和满足用户订阅。
实施例四
图9是本发明实施例四提供的一种话题订阅方法中的建立持续性话题与资
源的关联的流程图,本实施例在实施例三的基础上,对“建立所述持续性话题和所述资源库中的资源的关联,存入持续性话题检索库”进行了进一步优化,如图9所示,本实施例中的“建立所述持续性话题和所述资源库中的资源的关联,存入持续性话题检索库”具体包括如下:
S831,分别提取所述持续性话题和所述资源的词分布,所述持续性话题的统计语料包括所述持续性话题的点击资源和所述资源库中标题包含所述持续性话题的资源。
所述资源的词分布通过统计该资源的标题和正文中的词频来得到。
S832,根据所述持续性话题的词分布和词向量特征,对所述持续性话题进行去重和聚类处理。
根据所述持续性话题的词分布和词向量(word-embedding)特征,计算持续性话题之间的相似度,根据所述相似度对所述持续性话题进行去重和聚类处理。
其中,根据所述持续性话题的词分布和词向量特征,对所述持续性话题进行去重和聚类处理包括:
根据所述持续性话题的词分布和词向量特征,计算所述持续性话题之间的相似度;
保留所述相似度大于第一预设阈值的持续性话题中的一个持续性话题;
将所述相似度大于第二预设阈值且小于或等于第一预设阈值的持续性话题归为一类。
其中,所述第一预设阈值大于所述第二预设阈值。去重即将所述多个持续性话题的相似度大于第一预设阈值的持续性话题去掉,即去掉语义意思一致的持续性话题去掉,例如“跑男4”和“奔跑吧兄弟4”,只保留其一即可。聚类是指把语义意思相似的持续性话题归为一类,例如“王源照片”和“王源街拍”。通过持续性话题之间的相似度来对持续性话题进行去重和聚类处理,可以提高处理的准确度。
S833,建立处理后的持续性话题和资源的关联关系,存入持续性话题检索库。
建立进行去重和聚类处理后的持续性话题和资源的关联关系,并将持续性话题、资源和对应的关联关系存入持续性话题检索库。
其中,建立处理后的持续性话题和资源的关联关系,存入持续性话题检索库优选包括:
对所述处理后的持续性话题建立词分布的倒排索引;
根据资源的词分布和所述倒排索引,确定与该资源对应的候选持续性话题;
计算该资源正文的词分布和所述候选持续性话题的词分布的相似度;
将所述相似度大于预设阈值的候选持续性话题作为与该资源关联的持续性话题,存入持续性话题检索库。
其中,倒排索引(Inverted index),也常被称为反向索引、置入档案或反向档案,是一种索引方法,被用来存储在全文搜索下某个单词在一个文档或者一组文档中的存储位置的映射。通过倒排索引,可以根据单词快速获取包含这个单词的文档列表。
对处理后的持续性话题建立词分布的倒排索引,即确定词在持续性话题的位置,然后根据资源标题的词分布以及正文中出行次数较多的词,在所述倒排索引中检索出对应的持续性话题,作为候选持续性话题,该候选持续性话题包括至少一项,计算资源正文中的出行次数较多的词和该候选持续性话题的词分布之间的相似度,将所述相似度大于预设阈值的候选持续性话题作为与该资源关联的持续性话题,将所有的持续性话题与资源的关联关系确定后存入持续性话题检索库。通过上述方法建立持续性话题与资源的关联关联,能够更加及时的将新增的资源挂接到持续性话题下,进而展现到线上。
例如,标题为“太阳的后裔中的面膜”的资源,正文中多次提到了“面膜”、“补水”,这些词与持续性话题“太阳的后裔”的相似度就会非常低,故确定该资源与“太阳的后裔”不匹配;标题为“重温太阳的后裔的经典画面”的资源,正文中多次提到了“太阳的后裔”、“宋仲基”、“宋慧乔”,这些词跟持续性话题“太阳的后裔”的相关性都比较高,故确定该资源与“太阳的后裔”能够匹配。
S834,根据持续性话题关联的资源的词分布,对处理后的持续性话题关联的资源进行去重。
由于资源量比较大,直接在全量资源上对资源进行去重代价会比较高,因此,基于持续性话题和资源的匹配结果,对每个持续性话题所关联的资源进行去重。在对每个持续性话题所关联的资源进行去重时,主要考虑两个资源之间的资源标题的词分布、资源正文的词分布的相似度,来进行去重。
其中,根据持续性话题关联的资源的词分布,对处理后的持续性话题关联的资源进行去重优选包括:
采用流式比较的方式,当所述持续性话题关联的两个资源的相似度大于预设相似度阈值时,保留所述两个资源中价值得分较高的资源。
例如,已经保留了资源A、B、C,流式输入资源D,发现D跟A比较相似,且D的价值得分高于A的价值得分,则将资源D替换资源A,即保留资源D、B、C。通过流式比较的方式,可以提高比较速度。
本实施例的技术方案,通过分别提取持续性话题和资源的词分布,基于持续性话题和资源的词分布建立持续性话题与资源的关联关系,能够及时的将新增的资源挂接到持续性话题下,并及时推送给用户,通过在建立持续性话题与资源的关联关系后,对持续性话题关联的资源进行去重,可以提高去重的速度。
实施例五
图10是本发明实施例五提供的一种话题订阅装置的结构示意图,如图10所示,本实施例所述的话题订阅装置包括:话题匹配模块1010、话题显示模块1020和资源推荐模块1030。
其中,话题匹配模块1010用于根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;
话题推荐模块1020用于将所述持续性话题返回客户端显示,以供所述用户进行订阅;
资源推荐模块1030用于保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
优选的,还包括:
话题推荐模块,用于根据所述用户的历史行为及持续性话题的订阅行为,
确定向所述用户推荐的持续性话题并推荐给所述用户。
优选的,还包括:
匹配资源显示模块,用于在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示,其中,所述持续性话题检索库包括持续性话题和匹配资源的关联关系。
优选的,所述匹配资源显示模块包括:
匹配资源查询单元,用于在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源;
匹配资源排序单元,用于根据所述匹配资源的点击记录、资源属性和用户的历史行为,对所述匹配资源进行排序;
匹配资源推荐单元,用于按照排序结果,将所述匹配资源返回客户端显示。
优选的,所述话题匹配模块包括:
话题确定单元,根据用户的检索关键词,结合用户的历史行为和/或订阅记录,确定所述检索关键词的持续性话题;
话题排序单元,用于根据所述持续性话题的点击记录和用户的历史行为,对所述持续性话题进行排序。
优选的,还包括:
话题挖掘模块,用于在根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题之前,根据用户检索日志,挖掘出满足预设条件的用户查询作为持续性话题;
资源挖掘模块,用于根据预先配置网站和所述持续性话题的点击资源,建立资源库;
话题资源关联模块,用于建立所述持续性话题和所述资源库中的资源的关联,存入持续性话题检索库。
其中,所述预设条件优选包括:在第一预设天数内超过预设人数连续搜索第二预设天数,和/或包括预设关键词。
优选的,所述资源挖掘模块包括:
原始资源获取单元,用于获取所述预先配置网站中的资源和所述持续性话
题的点击资源,作为原始资源;
资源字面去重单元,用于根据所述原始资源的标题和正文,对字面内容一致的原始资源进行去重,得到资源;
资源属性提取单元,用于提取所述资源的资源属性,所述资源属性包括标题、摘要、配图和价值得分;
资源库建立单元,用于将所述资源和对应的资源属性存入资源库。
优选的,所述话题资源关联模块包括:
词分布提取单元,用于分别提取所述持续性话题和所述资源的词分布,所述持续性话题的统计语料包括所述持续性话题的点击资源和所述资源库中标题包含所述持续性话题的资源;
话题处理单元,用于根据所述持续性话题的词分布和词向量特征,对所述持续性话题进行去重和聚类处理;
话题资源关联单元,用于建立处理后的持续性话题和资源的关联关系,存入持续性话题检索库;
资源词分布去重单元,用于根据持续性话题关联的资源的词分布,对处理后的持续性话题关联的资源进行去重。
优选的,所述话题处理单元具体用于:
根据所述持续性话题的词分布和词向量特征,计算所述持续性话题之间的相似度;
保留所述相似度大于第一预设阈值的持续性话题中的一个持续性话题;
将所述相似度大于第二预设阈值且小于或等于第一预设阈值的持续性话题归为一类。
优选的,所述话题资源关联单元具体用于:
对所述处理后的持续性话题建立词分布的倒排索引;
根据资源的词分布和所述倒排索引,确定与该资源对应的候选持续性话题;
计算该资源正文的词分布和所述候选持续性话题的词分布的相似度;
将所述相似度大于预设阈值的候选持续性话题作为与该资源关联的持续性话题,存入持续性话题检索库。
优选的,所述资源词分布去重单元具体用于:
采用流式比较的方式,当所述持续性话题关联的两个资源的相似度大于预设相似度阈值时,保留所述两个资源中价值得分较高的资源。
上述产品可执行本发明任意实施例所提供的方法,具备执行方法相应的功能模块和有益效果。
实施例六
本发明实施例六提供了一种服务器,包括本发明任意实施例所提供的话题订阅装置。
具体的,如图11所示,本发明实施例提供一种服务器,该服务器包括处理器1110、存储器1120、输入装置1130和输出装置1140;服务器中处理器1110的数量可以是一个或多个,图11中以一个处理器1110为例;服务器中的处理器1110、存储器1120、输入装置1130和输出装置1140可以通过总线或其他方式连接,图11中以通过总线连接为例。
存储器1120作为一种计算机可读存储介质,可用于存储软件程序、计算机可执行程序以及模块,如本发明实施例中的话题订阅方法对应的程序指令/模块(例如,话题订阅装置中的话题匹配模块1010、话题显示模块1020和资源推荐模块1030)。处理器1110通过运行存储在存储器1120中的软件程序、指令以及模块,从而执行服务器的各种功能应用以及数据处理,即实现上述的话题订阅方法。
存储器1120可主要包括存储程序区和存储数据区,其中,存储程序区可存储操作系统、至少一个功能所需的应用程序;存储数据区可存储根据服务器的使用所创建的数据等。此外,存储器1120可以包括高速随机存取存储器,还可以包括非易失性存储器,例如至少一个磁盘存储器件、闪存器件、或其他非易失性固态存储器件。在一些实例中,存储器1120可进一步包括相对于处理器1110远程设置的存储器,这些远程存储器可以通过网络连接至服务器。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
输入装置1130可用于接收输入的数字或字符信息,以及产生与服务器的用户设置以及功能控制有关的键信号输入。输出装置1140可包括显示屏等显示设备。
本发明实施例还提供一种包含计算机可执行指令的存储介质,所述计算机可执行指令在由计算机处理器执行时用于执行一种话题订阅方法,该方法包括:
根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;
将所述持续性话题返回客户端显示,以供所述用户进行订阅;
保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
上述方案中,还可以包含其他计算机可执行指令,在计算机处理器执行时具体用于执行本发明实施例所提供的话题订阅方法的各个步骤。
通过以上关于实施方式的描述,所属领域的技术人员可以清楚地了解到,本发明可借助软件及必需的通用硬件来实现,当然也可以通过硬件实现,但很多情况下前者是更佳的实施方式。基于这样的理解,本发明的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品可以存储在计算机可读存储介质中,如计算机的软盘、只读存储器(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、闪存(FLASH)、硬盘或光盘等,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行本发明各个实施例所述的方法。
值得注意的是,上述话题更新装置的实施例中,所包括的各个单元和模块只是按照功能逻辑进行划分的,但并不局限于上述的划分,只要能够实现相应的功能即可;另外,各功能单元的具体名称也只是为了便于相互区分,并不用于限制本发明的保护范围。
以上所述,仅为本发明的具体实施方式,但本发明的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本发明披露的技术范围内,可轻易想到
的变化或替换,都应涵盖在本发明的保护范围之内。因此,本发明的保护范围应以所述权利要求的保护范围为准。
Claims (25)
- 一种话题订阅方法,其特征在于,所述方法包括:根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;将所述持续性话题返回客户端显示,以供所述用户进行订阅;保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
- 根据权利要求1所述的方法,其特征在于,还包括:根据所述用户的历史行为及持续性话题的订阅行为,确定向所述用户推荐的持续性话题并推荐给所述用户。
- 根据权利要求1所述的方法,其特征在于,还包括:在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示,其中,所述持续性话题检索库包括持续性话题和匹配资源的关联关系。
- 根据权利要求3所述的方法,其特征在于,在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示包括:在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源;根据所述匹配资源的点击记录、资源属性和用户的历史行为,对所述匹配资源进行排序;按照排序结果,将所述匹配资源返回客户端显示。
- 根据权利要求1所述的方法,其特征在于,根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题包括:根据用户的检索关键词,结合用户的历史行为和/或订阅记录,确定所述检索关键词的持续性话题;根据所述持续性话题的点击记录和用户的历史行为,对所述持续性话题进行排序。
- 根据权利要求1-5任一所述的方法,其特征在于,在根据用户的检索 关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题之前,还包括:根据用户检索日志,挖掘出满足预设条件的用户查询作为持续性话题;根据预先配置网站和所述持续性话题的点击资源,建立资源库;建立所述持续性话题和所述资源库中的资源的关联,存入持续性话题检索库。
- 根据权利要求6所述的方法,其特征在于,所述预设条件包括:在第一预设天数内超过预设人数连续搜索第二预设天数,和/或包括预设关键词。
- 根据权利要求6所述的方法,其特征在于,根据预先配置网站和所述持续性话题的点击资源,建立资源库包括:获取所述预先配置网站中的资源和所述持续性话题的点击资源,作为原始资源;根据所述原始资源的标题和正文,对字面内容一致的原始资源进行去重,得到资源;提取所述资源的资源属性,所述资源属性包括标题、摘要、配图和价值得分;将所述资源和对应的资源属性存入资源库。
- 根据权利要求6所述的方法,其特征在于,建立所述持续性话题和所述资源库中的资源的关联,存入持续性话题检索库包括:分别提取所述持续性话题和所述资源的词分布,所述持续性话题的统计语料包括所述持续性话题的点击资源和所述资源库中标题包含所述持续性话题的资源;根据所述持续性话题的词分布和词向量特征,对所述持续性话题进行去重和聚类处理;建立处理后的持续性话题和资源的关联关系,存入持续性话题检索库;根据持续性话题关联的资源的词分布,对处理后的持续性话题关联的资源进行去重。
- 根据权利要求9所述的方法,其特征在于,根据所述持续性话题的词 分布和词向量特征,对所述持续性话题进行去重和聚类处理包括:根据所述持续性话题的词分布和词向量特征,计算所述持续性话题之间的相似度;保留所述相似度大于第一预设阈值的持续性话题中的一个持续性话题;将所述相似度大于第二预设阈值且小于或等于第一预设阈值的持续性话题归为一类。
- 根据权利要求9所述的方法,其特征在于,建立处理后的持续性话题和资源的关联关系,存入持续性话题检索库包括:对所述处理后的持续性话题建立词分布的倒排索引;根据资源的词分布和所述倒排索引,确定与该资源对应的候选持续性话题;计算该资源正文的词分布和所述候选持续性话题的词分布的相似度;将所述相似度大于预设阈值的候选持续性话题作为与该资源关联的持续性话题,存入持续性话题检索库。
- 根据权利要求9所述的方法,其特征在于,根据持续性话题关联的资源的词分布,对处理后的持续性话题关联的资源进行去重包括:采用流式比较的方式,当所述持续性话题关联的两个资源的相似度大于预设相似度阈值时,保留所述两个资源中价值得分较高的资源。
- 一种话题订阅装置,其特征在于,所述装置包括:话题匹配模块,用于根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;话题显示模块,用于将所述持续性话题返回客户端显示,以供所述用户进行订阅;资源推荐模块,用于保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
- 根据权利要求13所述的装置,其特征在于,还包括:话题推荐模块,用于根据所述用户的历史行为及持续性话题的订阅行为,确定向所述用户推荐的持续性话题并推荐给所述用户。
- 根据权利要求13所述的装置,其特征在于,还包括:匹配资源显示模块,用于在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源,并返回客户端显示,其中,所述持续性话题检索库包括持续性话题和匹配资源的关联关系。
- 根据权利要求15所述的装置,其特征在于,所述匹配资源显示模块包括:匹配资源查询单元,用于在接收到用户点击的持续性话题时,在预先建立的持续性话题检索库中查询相应的匹配资源;匹配资源排序单元,用于根据所述匹配资源的点击记录、资源属性和用户的历史行为,对所述匹配资源进行排序;匹配资源推荐单元,用于按照排序结果,将所述匹配资源返回客户端显示。
- 根据权利要求13所述的装置,其特征在于,所述话题匹配模块包括:话题确定单元,用于根据用户的检索关键词,结合用户的历史行为和/或订阅记录,确定所述检索关键词的持续性话题;话题排序单元,用于根据所述持续性话题的点击记录和用户的历史行为,对所述持续性话题进行排序。
- 根据权利要求13-17任一所述的装置,其特征在于,还包括:话题挖掘模块,用于在根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题之前,根据用户检索日志,挖掘出满足预设条件的用户查询作为持续性话题;资源挖掘模块,用于根据预先配置网站和所述持续性话题的点击资源,建立资源库;话题资源关联模块,用于建立所述持续性话题和所述资源库中的资源的关联,存入持续性话题检索库。
- 根据权利要求18所述的装置,其特征在于,所述预设条件包括:在第一预设天数内超过预设人数连续搜索第二预设天数,和/或包括预设关键词。
- 根据权利要求18所述的装置,其特征在于,所述资源挖掘模块包括:原始资源获取单元,用于获取所述预先配置网站中的资源和所述持续性话题的点击资源,作为原始资源;资源字面去重单元,用于根据所述原始资源的标题和正文,对字面内容一致的原始资源进行去重,得到资源;资源属性提取单元,用于提取所述资源的资源属性,所述资源属性包括标题、摘要、配图和价值得分;资源库建立单元,用于将所述资源和对应的资源属性存入资源库。
- 根据权利要求18所述的装置,其特征在于,所述话题资源关联模块包括:词分布提取单元,用于分别提取所述持续性话题和所述资源的词分布,所述持续性话题的统计语料包括所述持续性话题的点击资源和所述资源库中标题包含所述持续性话题的资源;话题处理单元,用于根据所述持续性话题的词分布和词向量特征,对所述持续性话题进行去重和聚类处理;话题资源关联单元,用于建立处理后的持续性话题和资源的关联关系,存入持续性话题检索库;资源词分布去重单元,用于根据持续性话题关联的资源的词分布,对处理后的持续性话题关联的资源进行去重。
- 根据权利要求21所述的装置,其特征在于,所述话题处理单元具体用于:根据所述持续性话题的词分布和词向量特征,计算所述持续性话题之间的相似度;保留所述相似度大于第一预设阈值的持续性话题中的一个持续性话题;将所述相似度大于第二预设阈值且小于或等于第一预设阈值的持续性话题归为一类。
- 根据权利要求21所述的装置,其特征在于,所述话题资源关联单元具体用于:对所述处理后的持续性话题建立词分布的倒排索引;根据资源的词分布和所述倒排索引,确定与该资源对应的候选持续性话题;计算该资源正文的词分布和所述候选持续性话题的词分布的相似度;将所述相似度大于预设阈值的候选持续性话题作为与该资源关联的持续性话题,存入持续性话题检索库。
- 根据权利要求21所述的装置,其特征在于,所述资源词分布去重单元具体用于:采用流式比较的方式,当所述持续性话题关联的两个资源的相似度大于预设相似度阈值时,保留所述两个资源中价值得分较高的资源。
- 一种包含计算机可执行指令的存储介质,所述计算机可执行指令在由计算机处理器执行时用于执行一种话题订阅方法,其特征在于,该方法包括:根据用户的检索关键词,结合用户的历史行为和/或订阅记录,为所述检索关键词匹配持续性话题;将所述持续性话题返回客户端显示,以供所述用户进行订阅;保存所述用户订阅的持续性话题,并在所述用户订阅的持续性话题对应的匹配资源有更新时,将更新的匹配资源推荐给用户。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US16/234,337 US11429680B2 (en) | 2016-06-27 | 2018-12-27 | Topic subscription method and apparatus, and storage medium |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610481888.4 | 2016-06-27 | ||
CN201610481888.4A CN105930539A (zh) | 2016-06-27 | 2016-06-27 | 话题订阅方法和装置 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US16/234,337 Continuation US11429680B2 (en) | 2016-06-27 | 2018-12-27 | Topic subscription method and apparatus, and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2018000569A1 true WO2018000569A1 (zh) | 2018-01-04 |
Family
ID=56828203
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2016/097477 WO2018000569A1 (zh) | 2016-06-27 | 2016-08-31 | 话题订阅方法、装置和存储介质 |
Country Status (3)
Country | Link |
---|---|
US (1) | US11429680B2 (zh) |
CN (1) | CN105930539A (zh) |
WO (1) | WO2018000569A1 (zh) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110083774A (zh) * | 2019-05-10 | 2019-08-02 | 腾讯科技(深圳)有限公司 | 应用推荐列表的确定方法、装置、计算机设备及存储介质 |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105930539A (zh) | 2016-06-27 | 2016-09-07 | 北京百度网讯科技有限公司 | 话题订阅方法和装置 |
CN107122397A (zh) * | 2017-03-15 | 2017-09-01 | 百度在线网络技术(北京)有限公司 | 内容推荐方法及装置 |
CN107346336B (zh) | 2017-06-29 | 2021-06-08 | 北京百度网讯科技有限公司 | 基于人工智能的信息处理方法和装置 |
CN108305051A (zh) * | 2018-02-09 | 2018-07-20 | 网易(杭州)网络有限公司 | 邮件展示方法、介质、系统和电子设备 |
CN110472021A (zh) * | 2018-05-11 | 2019-11-19 | 微软技术许可有限责任公司 | 会话中推荐新闻的技术 |
CN109106352A (zh) * | 2018-08-30 | 2019-01-01 | 缤刻普达(北京)科技有限责任公司 | 血压监控方法、系统及血压计 |
CN109582852B (zh) * | 2018-12-05 | 2021-04-09 | 中国银行股份有限公司 | 一种全文检索结果的排序方法及系统 |
CN110737694B (zh) * | 2019-09-17 | 2022-09-23 | 中电万维信息技术有限责任公司 | 一种政务数据资源目录智能订阅方法 |
CN112784147B (zh) * | 2019-11-04 | 2024-09-27 | 阿里巴巴(深圳)技术有限公司 | 信息处理方法、装置、设备及系统 |
CN110955816B (zh) * | 2019-11-08 | 2022-11-08 | 广州坚和网络科技有限公司 | 基于内容标签的聚合主题内容的方法 |
CN111680152B (zh) * | 2020-06-10 | 2023-04-18 | 创新奇智(成都)科技有限公司 | 目标文本的摘要提取方法及装置、电子设备、存储介质 |
CN112464084B (zh) * | 2020-11-17 | 2021-07-02 | 厦门立马耀网络科技有限公司 | 基于大数据定位和人工智能的业务优化方法及云计算中心 |
CN112765464A (zh) * | 2021-01-13 | 2021-05-07 | 中国地质大学(武汉) | 面向主题的新内容检索推送方法、装置、设备及存储介质 |
CN116204664A (zh) * | 2021-12-01 | 2023-06-02 | 北京字节跳动网络技术有限公司 | 一种内容订阅方法、装置、计算机设备及存储介质 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100057716A1 (en) * | 2008-08-28 | 2010-03-04 | Stefik Mark J | System And Method For Providing A Topic-Directed Search |
CN103577501A (zh) * | 2012-08-10 | 2014-02-12 | 深圳市世纪光速信息技术有限公司 | 热点话题搜索系统及热点话题搜索方法 |
CN103577478A (zh) * | 2012-08-06 | 2014-02-12 | 腾讯科技(深圳)有限公司 | 网页推送方法及系统 |
CN105930539A (zh) * | 2016-06-27 | 2016-09-07 | 北京百度网讯科技有限公司 | 话题订阅方法和装置 |
Family Cites Families (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10635723B2 (en) * | 2004-02-15 | 2020-04-28 | Google Llc | Search engines and systems with handheld document data capture devices |
US8539359B2 (en) * | 2009-02-11 | 2013-09-17 | Jeffrey A. Rapaport | Social network driven indexing system for instantly clustering people with concurrent focus on same topic into on-topic chat rooms and/or for generating on-topic search results tailored to user preferences regarding topic |
CN102129452A (zh) * | 2011-03-04 | 2011-07-20 | 李丽平 | 未来信息获取系统及获取方法 |
CN103177024A (zh) * | 2011-12-23 | 2013-06-26 | 微梦创科网络科技(中国)有限公司 | 一种话题信息展现方法和装置 |
WO2014032014A2 (en) * | 2012-08-24 | 2014-02-27 | Kao Chia Lung | Cumulative encyclopedia method, system and device |
CN102945244A (zh) * | 2012-09-24 | 2013-02-27 | 南京大学 | 基于句号特征字串的中文网页重复文档检测和过滤方法 |
US10228819B2 (en) * | 2013-02-04 | 2019-03-12 | 602531 British Cilumbia Ltd. | Method, system, and apparatus for executing an action related to user selection |
US20140280241A1 (en) * | 2013-03-15 | 2014-09-18 | MediaGraph, LLC | Methods and Systems to Organize Media Items According to Similarity |
US9767099B2 (en) * | 2013-09-12 | 2017-09-19 | Zappylab, Inc. | System and method for dynamic interaction with a research publication database |
CN103593418B (zh) * | 2013-10-30 | 2017-03-29 | 中国科学院计算技术研究所 | 一种面向大数据的分布式主题发现方法及系统 |
CN104636402B (zh) * | 2013-11-13 | 2018-05-01 | 阿里巴巴集团控股有限公司 | 一种业务对象的分类、搜索、推送方法和系统 |
CN104869439B (zh) * | 2015-05-14 | 2018-03-09 | 无锡天脉聚源传媒科技有限公司 | 一种视频推送方法及装置 |
-
2016
- 2016-06-27 CN CN201610481888.4A patent/CN105930539A/zh active Pending
- 2016-08-31 WO PCT/CN2016/097477 patent/WO2018000569A1/zh active Application Filing
-
2018
- 2018-12-27 US US16/234,337 patent/US11429680B2/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100057716A1 (en) * | 2008-08-28 | 2010-03-04 | Stefik Mark J | System And Method For Providing A Topic-Directed Search |
CN103577478A (zh) * | 2012-08-06 | 2014-02-12 | 腾讯科技(深圳)有限公司 | 网页推送方法及系统 |
CN103577501A (zh) * | 2012-08-10 | 2014-02-12 | 深圳市世纪光速信息技术有限公司 | 热点话题搜索系统及热点话题搜索方法 |
CN105930539A (zh) * | 2016-06-27 | 2016-09-07 | 北京百度网讯科技有限公司 | 话题订阅方法和装置 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110083774A (zh) * | 2019-05-10 | 2019-08-02 | 腾讯科技(深圳)有限公司 | 应用推荐列表的确定方法、装置、计算机设备及存储介质 |
CN110083774B (zh) * | 2019-05-10 | 2023-11-03 | 腾讯科技(深圳)有限公司 | 应用推荐列表的确定方法、装置、计算机设备及存储介质 |
Also Published As
Publication number | Publication date |
---|---|
US20190155856A1 (en) | 2019-05-23 |
CN105930539A (zh) | 2016-09-07 |
US11429680B2 (en) | 2022-08-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2018000569A1 (zh) | 话题订阅方法、装置和存储介质 | |
CN109800352B (zh) | 基于剪贴板进行信息推送的方法、系统及终端设备 | |
US8341147B2 (en) | Blending mobile search results | |
US7860878B2 (en) | Prioritizing media assets for publication | |
CN107660284B (zh) | 基于机器学习的搜索改进 | |
WO2017005054A1 (zh) | 视频推荐方法、服务器和存储介质 | |
JP6538277B2 (ja) | 検索クエリ間におけるクエリパターンおよび関連する総統計の特定 | |
US20150262069A1 (en) | Automatic topic and interest based content recommendation system for mobile devices | |
WO2016150083A1 (zh) | 一种信息输入方法和装置 | |
KR102001647B1 (ko) | 지식 패널들을 컨텍스트화 | |
US10552422B2 (en) | Extended search method and apparatus | |
CN107766399B (zh) | 用于使图像与内容项目匹配的方法和系统及机器可读介质 | |
CN103106287B (zh) | 一种用户检索语句的处理方法及系统 | |
WO2014056337A1 (zh) | 搜索词获取方法、服务器、搜索词推荐系统 | |
WO2017143930A1 (zh) | 一种搜索结果排序方法及其设备 | |
KR20140032439A (ko) | 전자 디바이스에 근접하여 현재 디스플레이되고 있는 텔레비전 프로그램을 결정함으로써 사용자 검색 결과들을 향상시키기 위한 시스템 및 방법 | |
CN107463592B (zh) | 用于将内容项目与图像匹配的方法、设备和数据处理系统 | |
CN107491465B (zh) | 用于搜索内容的方法和装置以及数据处理系统 | |
JP5952711B2 (ja) | 予測対象コンテンツにおける将来的なコメント数を予測する予測サーバ、プログラム及び方法 | |
US20150052126A1 (en) | Method and system for recommending relevant web content to second screen application users | |
US9824149B2 (en) | Opportunistically solving search use cases | |
EP3144825A1 (en) | Enhanced digital media indexing and retrieval | |
US20170075999A1 (en) | Enhanced digital media indexing and retrieval | |
CN107784061B (zh) | 确定基于图像的内容样式的方法和系统及机器可读介质 | |
Hirata et al. | A web agent based on exploratory event mining in social media |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 16906966 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 16906966 Country of ref document: EP Kind code of ref document: A1 |