WO2015101155A1 - 向用户推荐信息的方法 - Google Patents

向用户推荐信息的方法 Download PDF

Info

Publication number
WO2015101155A1
WO2015101155A1 PCT/CN2014/093660 CN2014093660W WO2015101155A1 WO 2015101155 A1 WO2015101155 A1 WO 2015101155A1 CN 2014093660 W CN2014093660 W CN 2014093660W WO 2015101155 A1 WO2015101155 A1 WO 2015101155A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
information
keyword
words
candidate recommendation
Prior art date
Application number
PCT/CN2014/093660
Other languages
English (en)
French (fr)
Inventor
江潮
Original Assignee
语联网(武汉)信息技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority to US14/779,240 priority Critical patent/US20160321543A1/en
Application filed by 语联网(武汉)信息技术有限公司 filed Critical 语联网(武汉)信息技术有限公司
Publication of WO2015101155A1 publication Critical patent/WO2015101155A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00Computing arrangements using knowledge-based models
    • G06N5/02Knowledge representation; Symbolic representation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/245Query processing
    • G06F16/2458Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2465Query processing support for facilitating data mining operations in structured databases
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/40Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
    • G06F16/43Querying
    • G06F16/435Filtering based on additional data, e.g. user or group profiles
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/903Querying
    • G06F16/90335Query processing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/953Querying, e.g. by the use of web search engines
    • G06F16/9535Search customisation based on user profiles and personalisation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q20/00Payment architectures, schemes or protocols
    • G06Q20/08Payment architectures
    • G06Q20/12Payment architectures specially adapted for electronic shopping systems
    • G06Q20/123Shopping for digital content
    • G06Q20/1235Shopping for digital content with control of digital rights management [DRM]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q20/00Payment architectures, schemes or protocols
    • G06Q20/38Payment protocols; Details thereof
    • G06Q20/40Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • G06Q20/401Transaction verification
    • G06Q20/4016Transaction verification involving fraud or risk level assessment in transaction processing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q20/00Payment architectures, schemes or protocols
    • G06Q20/38Payment protocols; Details thereof
    • G06Q20/40Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • G06Q20/405Establishing or using transaction specific rules
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • G06N20/20Ensemble learning

Definitions

  • the invention belongs to the field of streaming media information processing and text information retrieval, and in particular relates to a method for recommending information to a user.
  • streaming media information processing mainly focuses on achieving efficiency, while text information retrieval is mainly used to search for related information, and has not seen the work of closely combining the two.
  • Mobile text mainly refers to text information that flows continuously over time.
  • Mobile text is generally closely related to streaming media playback, such as video subtitles, scene descriptions, audio lyrics and interpretations, as well as various other textual content corresponding to audio, video, and other content representations that change over time. At present, this part of the information is rare except for movie subtitles and lyrics. Users often use streaming information to promote interest and want to know more about the associated information. Flowing text can assist the user in specifying key points of interest.
  • User preferences indicate what information the user is interested in. They are mainly used in systems like Amazon and Taobao. They can be based on user information for product recommendation, or similar interest-based recommendations in social networks, and search engine-assisted search information. search for.
  • the user preference information can be set according to the personal preference setting interface provided, and the data accessed by the user can be recorded and analyzed, and the methods of statistical analysis, machine learning, and the like can be used for learning and sorting.
  • the object of the present invention is to provide a method for information processing, which can assist the user to quickly obtain the related information recommendation of the object being viewed, and improve the recommendation. Accuracy and efficiency.
  • a brief summary is given below. This generalization is not a general comment, nor is it intended to identify key/critical constituent elements or to depict these embodiments. The scope of protection. Its sole purpose is to present some concepts in a simplified form as a prelude to the following detailed description.
  • the present invention provides a method for recommending information to a user, including:
  • the process of obtaining an initial keyword for viewing a user by the user includes:
  • the description information of the viewing object and the information in the current text window are displayed to the user, and the user specifies a word as the initial keyword.
  • the associated database comprises:
  • the user preference information can be obtained by:
  • the user preference information selected by the user is accepted through the software interface, or the user preference information is generated by recording and analyzing the data accessed by the user.
  • the method further includes: if the cardinality of the candidate recommendation information set is less than a quantity threshold, performing the association expansion step by using the elements in the candidate recommendation information set as the keywords one by one, and adding the generated related words To the candidate recommendation information set, the elements in the set are filtered according to the user preference information.
  • the method further includes: after determining that the cardinality of the candidate recommendation information set is not less than a quantity threshold or the number of times the performing the extension step is performed is greater than a number of times threshold, then all of the candidate recommendation information sets except the initial keyword are output. The word ends.
  • the step of associating the extension is performed one or more times.
  • the auxiliary user quickly obtains the recommendation of the related information of the object being viewed
  • the information recommended to the user is generated through the association database and user preferences, and both can improve the accuracy of the recommendation information;
  • the user can directly expand the association by directly specifying the keyword.
  • the process of opening the search engine, inputting the keyword, searching and then selecting the result can be directly extended through the recommendation information to improve the information browsing efficiency.
  • Figure 1 is a schematic flow diagram of some illustrative embodiments
  • FIG. 2 is a schematic diagram of a recommended flow in some illustrative embodiments
  • FIG. 3 is a schematic diagram of an associated database in some illustrative embodiments.
  • FIG. 4 is a schematic diagram of a related network of an associated database in some illustrative embodiments.
  • Figure 5 is a schematic illustration of an information recommendation system in some illustrative embodiments.
  • Figure 6 is a schematic illustration of user preferences in some illustrative embodiments.
  • Figure 7 is a schematic diagram showing the expansion of the secondary association information in some illustrative embodiments.
  • Figure 8 is a schematic illustration of the continuous expansion of associated information in some illustrative embodiments.
  • the original media to which the present invention relates may be video, audio, text, and the like.
  • the present invention provides a method for recommending information to a user, which specifically includes the following steps:
  • a process of quickly recommending information to the user is realized. For example, if the keyword currently selected by the user is an actor, other information related to the actor can be obtained, and the actor profile and related works that are seen can be quickly understood; if the user likes the clothing, the introduction of the clothing can be quickly understood. Even if you like the attractions, you can quickly get the introduction of the attraction, or even book the tickets; if the user likes the animals, you can quickly get an introduction to the animal, which related documentary, know that the zoo can watch.
  • Figure 2 shows a schematic diagram of the recommended flow in some illustrative embodiments
  • the viewed object information may be information description content and any text information in the text window, wherein the original media may be any content that the user is watching and listening, such as video and audio. , text, and so on.
  • the media information there may be some descriptions of the basic information.
  • a movie generally has a director, a producer, a main actor, a brief introduction, a related introduction, and the like.
  • the description of the information is such an overall description of the original media.
  • the content in the text window is the information in the original media that the user can directly observe. Still taking movies as an example, for any movie scene, there are people, events, places, backgrounds, etc., especially some additional information in pictures or lines, such as attractions, costumes, citations, and some history, culture, etc. . These content are constantly changing over time and can be used as content in a text window. That is, the specific content in the scene of interest when the user pauses. Receiving a trigger operation of the user selecting one of the words based on the object information viewed by the user, that is, the information describing the content and the text information of the text window; for example, a word selected by the user when pausing.
  • the candidate recommendation information set is initialized to be empty.
  • the word selected by the user is used as an initial keyword and added to the candidate recommendation information set.
  • the user preference information is personal preference information set by the user. For example, users may have a greater interest in people, history, attractions, or fashion information, and so on.
  • the user preference representation is based on the following content representation and association structure, and is mainly used to describe the user's interest in which relation to Content. For example, the user may be interested in only the part of the video that is only news and documentary, the movie is only interested in the director and the starring work, the historical architectural background appearing in the movie scene, the clothing brand of an actor is even wearing Interested in costumes.
  • the user preference information is represented in the form of FIG. 6, where Content is the media object to be described, and relation i is the related information item to be filled out.
  • the related words associated with the film are: ⁇ movie: producer, director, starring, music, type,...>;
  • the related words of the starring association are: ⁇ starring: generic, works,...>;
  • the user preferences collection F ⁇ user preferences-movie, user preferences-zoo, movie-director, movie-starring, starring-genus, starring-works,... ⁇
  • the user views the data as the host: Po
  • the protagonist: Po is the initialization keyword selected by the user
  • the candidate recommendation information set Q ⁇ starring: Po ⁇ ;
  • the initial candidate recommendation information set is empty, and the initial keyword is added.
  • FIG. 3 is a schematic diagram of an associated database using the movie "Kung Fu Panda" as an example in some illustrative embodiments; the associated database stores or retrieves associated data from the Internet, through the following Way of representation.
  • Content 2 Content 1 may be in the relation i
  • Content 1 may be Content relation j 2.
  • Movie Kung Fu Panda meets user preferences - movie; starring: Po, starring: Monkey, starring: Tigress is in line with the movie - starring; generic: giant panda in line with starring - generic;
  • the base of Q in S26 is five, which is smaller than the minimum recommended number threshold of six in the embodiment, and the "cardinality" is expressed as the number of information in the set.
  • the elements in the candidate recommendation information set need to be used as the keyword one by one, the association expansion step is performed, and the generated related words are added to the candidate recommendation information set. Filter the elements in the collection based on user preference information.
  • the associated information expansion step continues.
  • the elements in Q are taken one by one to expand the associated information, and all the extended results are added to Q.
  • each element in the candidate recommendation set is used as a keyword one by one, and the candidate recommendation information set obtained after being re-associated.
  • the recommendation information is Q except for the initial keyword "starring: Po", which is ⁇ movie: Kung Fu Panda, starring: Monkey, starring: Tigress, generic: giant panda, zoo: Wuhan Zoo, zoo: Atlanta Zoo ⁇ .
  • the associated extension step in the above embodiment may be performed one or more times as needed, and adjusted by the set number of associated times threshold. For example, set to 2 times.
  • Expanding 1 association can recommend:
  • Extending the association 2 times can recommend:
  • FIG. 5 shows a schematic diagram of an information recommendation system in some illustrative embodiments.
  • the viewed object may be any information in the information description and the text window.
  • the information description in the figure is some general description information about any content of the original media that the user is watching and listening, and can provide continuous information for the user to select; for example, video, audio, text, etc., for the media information
  • the text window in the figure is the information in the original media that the user can directly observe, that is, the word information at a certain moment. Still taking movies as an example, for any movie scene, there are people, events, places, backgrounds, etc., especially some additional information in pictures or lines, such as attractions, costumes, citations, and some history, culture, and so on. These content are constantly changing over time and can be used as content in a text window. That is, the specific content of the pause moment in the scene of interest when the user pauses.
  • the information description and the text window are the mobile information text representation of the original media, and the main purpose of the mobile information text representation is to facilitate the user who uses the original media to quickly associate the viewed information to the content of interest, the content of interest. It usually comes from what appears in the current short time window. Whether it is an information description or a text window, the content can be expressed as a description of the content surrounding the current information, and the description includes various components, and may also include various related information.
  • All the flow information texts represent the description object as the initial, the related information is gradually extended, and the related information constitutes a loose set, wherein the data correlation is mainly extended by the alternative conversion of Content and relation, that is, the flow information text representation It is a collection of Content, and the content in a Content can be the Content of other elements in the collection. Through this conversion relationship, you can expand the user's viewing part or listen to parts to as many aspects as possible.
  • a data-related network with associated and associated words as basic constituent elements, and the data-related networks are expanded by the same data layer and associated with each other.
  • the user preference is the personal preference information set by the user, and the collection F storage user's preference information mentioned in the embodiment, that is, the collection F of the object information that the user pays attention to, can directly provide the user selection, or collect information from the user. .
  • the auxiliary user can quickly obtain the related information recommendation of the object being viewed, improve the accuracy, reduce the disturbance, and improve the efficiency; when the user determines the content point of interest through the selection, the related information can be quickly provided for the user to select. Reduce the trouble of user manual input, improve the accuracy and efficiency of users to obtain content of interest.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • Accounting & Taxation (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Finance (AREA)
  • General Business, Economics & Management (AREA)
  • Computer Security & Cryptography (AREA)
  • Strategic Management (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Computer Hardware Design (AREA)
  • Fuzzy Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Multimedia (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Debugging And Monitoring (AREA)

Abstract

本发明公开了一种向用户推荐信息的方法,包括:获取用户查看对象的初始关键词,加入候选推荐信息集合;执行以下关联扩展步骤:从候选推荐信息集合中逐一获取元素作为关键词,在关联数据库中,查找与所述关键词相匹配的中心词,得到与所述中心词相关联的所有关联词;将得到的所有关联词加入候选推荐信息集合,根据用户喜好信息对集合中的元素进行过滤;所述关联扩展步骤可以根据需要一次或多次执行;向用户输出候选推荐信息集合中除初始关键词外的所有词。采用上述方案,能够帮助用户迅速获取感兴趣的信息,减少用户需要手动输入和人为筛选的麻烦,提高用户获取感兴趣内容的准确率和效率。

Description

向用户推荐信息的方法 技术领域
本发明属于流媒体信息处理和文本信息检索领域,尤其涉及向用户推荐信息的方法。
背景技术
目前流媒体信息处理主要集中在实现效率方面,而文本信息检索主要用于搜索相关信息,尚未见到将两者进行紧密结合的工作。流动文本,主要指随着时间不断流动的文本信息。流动文本一般和流媒体播放紧密相关,如视频的字幕、场景描述,音频的歌词和阐释,还有其它各种以音频、视频等随时间而变化的内容表现形式对应的文本内容。目前,这部分信息除了电影字幕和歌词之外还很少。用户经常会在使用流媒体信息时,促发兴趣想进一步了解由此引发的关联信息。流动文本可以辅助用户指定感兴趣的关键点。
用户喜好说明用户对哪些信息感兴趣,主要用于类似于Amazon、淘宝网等系统中,可以基于用户信息进行产品推荐,或者社交网络中基于相似兴趣的推荐,以及搜索引擎中利用历史搜索信息辅助搜索。用户喜好信息可以根据提供的个人喜好设置界面进行设置,也可以记录和分析用户访问的数据,使用统计分析、机器学习等方法进行学习和整理。
有一些系统能够对视频、音频进行推荐,都属于对相同类型、类似数据的推荐,和本发明要解决的问题不同。
所述这些方案,都不能让用户获得正在查看对象的感兴趣关联信息推荐。
发明内容
有鉴于此,为解决用户不能快速获取查看对象的感兴趣关联信息推荐的问题,本发明的目的是提出一种信息处理的方法,能够辅助用户快速获得正在查看对象的感兴趣关联信息推荐,提高准确性和效率。为了对披露的实施例的一些方面有一个基本的理解,下面给出了简单的概括。该概括部分不是泛泛评述,也不是要确定关键/重要组成元素或描绘这些实施例 的保护范围。其唯一目的是用简单的形式呈现一些概念,以此作为后面的详细说明的序言。
本发明提供一种向用户推荐信息的方法,包括:
获取用户查看对象的初始关键词,加入候选推荐信息集合;
执行以下关联扩展步骤:
从候选推荐信息集合中逐一获取元素作为关键词,在关联数据库中,查找与所述关键词相匹配的中心词,得到与所述中心词相关联的所有关联词;
将得到的所有关联词加入候选推荐信息集合,根据用户喜好信息对集合中的元素进行过滤;
向用户输出候选推荐信息集合中除初始关键词外的所有词。
优选地,所述获取用户查看对象的初始关键词的过程包括:
向用户显示查看对象的描述信息和当前文本窗口中的信息,由用户指定一个词作为初始关键词。
优选地,所述关联数据库包括:
多个中心词,以及与所述每个中心词相对应的至少一个关联词。
优选地,所述用户喜好信息可以通过以下方式获得:
通过软件界面接受用户勾选的用户喜好信息,或者通过记录和分析用户访问的数据,生成用户喜好信息。
优选地,还包括:如果所述候选推荐信息集合的基数小于数量阈值,则将所述候选推荐信息集合中的元素逐个作为所述关键词,执行所述关联扩展步骤,并将产生的关联词添加到所述候选推荐信息集合,根据用户喜好信息对集合中的元素进行过滤。
优选地,还包括:在判断所述候选推荐信息集合的基数不小于数量阈值或执行所述关联扩展步骤的次数大于次数阈值,则在输出所述候选推荐信息集合中除初始关键词外的所有词之后结束。
优选地,所述关联扩展的步骤执行一次或多次。
采用上述实施例,具有以下的有益效果:
1、辅助用户快速获得正在查看对象的感兴趣关联信息推荐;
2、推荐给用户的信息是通过关联数据库和用户喜好产生的,两者可以提高推荐信息的准确性;
3、用户可以通过直接指定关键词进行关联扩展,不需要通过打开搜索引擎,输入关键词,搜索然后选择结果的过程,可以直接通过推荐信息不断进行关联扩展,提高信息浏览效率。
为了上述以及相关的目的,一个或多个实施例包括后面将详细说明并在权利要求中特别指出的特征。下面的说明以及附图详细说明某些示例性方面,并且其指示的仅仅是各个实施例可以利用的各种方式中的一些方式。其它的益处和新颖性特征将随着下面的详细说明结合附图考虑而变得明显,所公开的实施例是要包括所有这些方面以及它们的等同。
说明书附图
图1是一些说明性实施例中的流程示意图;
图2是一些说明性实施例中的推荐流程示意图;
图3是一些说明性实施例中的关联数据库示意图;
图4是一些说明性实施例中的关联数据库的相关网络示意图;
图5是一些说明性实施例中的信息推荐系统示意图;
图6是一些说明性实施例中用户喜好表示的示意图;
图7是一些说明性实施例中的2次关联信息扩展示意图;
图8是一些说明性实施例中的连续进行关联信息扩展示意图。
具体实施方式
以下结合附图和实施例,对本发明进行进一步详细说明。应当理解,所描述的具体实施例仅仅用以解释本发明,并不限定本发明。
本发明涉及到的原始媒体可以是视频、音频、文本等等。
如图1所示,本发明提供一种向用户推荐信息的方法,具体包括如下步骤:
S11、获取用户查看对象的初始关键词,加入候选推荐信息集合;
执行以下关联扩展步骤:
S12、从候选推荐信息集合中逐一获取元素作为关键词,在关联数据库中,查找与所述关键词相匹配的中心词,得到与所述中心词相关联的所有关联词;
S13、将得到的所有关联词加入候选推荐信息集合,根据用户喜好信息对集合中的元素进行过滤;
S14、向用户输出候选推荐信息集合中除初始关键词外的所有词。
通过上述步骤,实现快捷的向用户推荐信息的过程。例如,用户当前选择的关键词为演员,则可得到与该演员相关的其它信息,可以快速了解看到的演员简介及其相关作品;如果用户喜欢其中的服饰,可以快速了解该服饰的介绍,甚至购买方式;如果用户喜欢其中的景点,可以快速获取景点介绍,甚至预订门票;如果用户喜欢其中的动物,可以快速获取该动物的介绍,有哪些相关的纪录片,知道那个动物园可以观看。以收听歌曲,如果用户喜欢曲风,可以快速了解作曲人,还有那些作品,哪些是经典作品;如果用户喜欢词,可以快速了解作词人,其它作品,经典作品;如果用户喜欢曲风,可以迅速了解该曲风的来源、发展,及该曲风最初始的歌曲,经典的歌曲等;如果用户喜欢演唱者,可以快速了解演唱者的作品和经典作品,专辑、演唱会,其他作品,即将推出的演唱会或作品,并预定门票等。
以下提供一个优选的实施例对上述步骤作进一步的详细说明:
如图2示出一些说明性实施例中的推荐流程示意图;
用户正在收看、收听原始媒体的任何内容的时候,查看到的对象信息可以是信息描述内容和文本窗口中的任何文本信息,其中原始媒体可以是用户正在收看、收听的任何内容,例如视频、音频、文本、等等。对该媒体信息,可以有一些基础信息的描述,以电影为例,一般有导演、制片、主要演员、内容简介、相关介绍、等等。信息描述就是这样一些对原始媒体的整体描述内容。
文本窗口中的内容则是用户能够直接观察到的原始媒体中的信息。还是以电影为例,对于任何一个电影场景,都有人物、事件、地点、背景、等等,特别是画面或台词中的一些附加信息,例如景点、服饰、引用、以及一些历史、文化等等。这些内容都是随着时间不断变化的,可以作为文本窗口中的内容。也就是当用户暂停时,所关注场景中的具体内容。基于用户查看的对象信息,即信息描述内容和文本窗口的文本信息,接收用户选择其中某个词的触发操作;例如用户通过暂停时选择的某个词。
候选推荐信息集合初始化为空。将用户选择的词作为初始关键词,加入候选推荐信息集合中。
下面结合电影《功夫熊猫》和图3所示的关联数据库的示意图,对图2 作进一步详细说明。
S21:初始化用户喜好信息的集合F;
用户喜好信息是用户设定的个人偏好信息。例如用户可能对人物、历史、景点或时尚信息有比较大的兴趣,等等。
用户喜好表示是基于以下的内容表示和关联的结构,主要用于描述用户对Content的哪些relation感兴趣。例如用户可能对视频中仅仅是新闻和纪录片的部分感兴趣,对电影则仅仅是导演和主演的作品感兴趣,对电影场景中出现的历史建筑背景,对某个演员的服饰品牌甚至就是穿着的服饰感兴趣。
<Content:relation1,relation2,...>
例如,以图6的形式表示用户喜好的信息,其中,Content为要说明的媒体对象,relationi为需要填写的关联信息项目。
例如:
电影关联的关联词为:<电影:制片人,导演,主演,音乐,类型,...>;
主演关联的关联词为:<主演:类属,作品,...>;
在实施例中,用户喜好集合F={用户喜好-电影,用户喜好-动物园,电影-导演,电影-主演,主演-类属,主演-作品,...}
S22:根据用户指定的初始关键词,初始化候选推荐信息集合Q;
例如,用户查看数据为主演:Po,主演:Po为用户选择的初始化关键词,候选推荐信息集合Q={主演:Po};
初始化候选推荐信息集合为空,加入初始关键词。
S23:扩展初始化次数为0;
将扩展次数设置为0,即extimes=0;
S24:依据关联信息中的数据关联信息扩展Q;
以主演:Po为中心词在关联数据库中进行关联扩展,即对关联集合Q进行首次关联扩展之后,得到关联后的关联集合Q={主演:Po,电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫,编剧:JonathanAibel}。
关联数据库,可参见图3、图4所示的关联关系。
如图3,是一些说明性实施例中以电影《功夫熊猫》为例的关联数据库示意图;所述的关联数据库存储或从互联网检索生成关联数据,通过以下 方式表示。
对于任何一个词,都会有很多相关的词。以一个词为中心,将所有发现的相关词与其进行关联,可以表示为:
<Content1:relation1,relation2,...relationi,...>
<Content2:relation3,relation4,...relationj,...>
......
其中,Content2可以是Content1中的relationi,Content1也可以是Content2中的relationj
例如,以电影:功夫熊猫作为中心词,可以有这样的一些相关的关联词,<电影:功夫熊猫:上映时间:2008年,制作人:Jonathan Aibel,导演:John Steffensen,译制:上海电影译制厂,主演:Po,......>,对于其中的关联词“主演:Po”又可以作为中心词,可以有这样的一些相关的关联词:<主演:Po:电影:功夫熊猫,主演:Monkey,主演:Tigross,类属:大熊猫,编剧:Jonathan Aibel,......>,将其中的关联词“类属:大熊猫”作为中心词,可以有这样的一些相关的关联词:<类属:大熊猫:主演:Po,分布区域:中国西南地区,外形:丰腴富态,头圆尾短,毛色黑白相间分明,动物园:武汉动物园,动物园:亚特南大动物园,......>,如此,对所有能够获得的数据,进行这种关联和关联词的表示,就可以产生以关联词为基本组成元素的数据相关网络。
如图4所示,将上述功夫熊猫到大熊猫的例子合成在一起,就形成了一个简单的以关联和关联词为基本组成元素的数据相关网络,用曲线关联的数据表示同一个数据。
S25:扩展次数加1;即extimes=0+1;
S26:根据F对Q进行过滤;
将得到的所有关联词加入候选推荐信息集合,根据用户喜好信息集合F对集合Q中的元素进行过滤;
将以Po为中心词在关联数据库中进行关联扩展,得到的候选推荐信息集合Q={主演:Po,电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫,编剧:Jonathan Aibel}。采用用户喜好的信息集合F对集合Q进行过滤,
定义Q=F∩Q,其中第一个Q为操作结果,根据用户喜好的集合F对 候选推荐信息集合Q进行过滤;
Q=F∩Q
={主演:Po,电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫}
其中:电影:功夫熊猫符合用户喜好-电影;主演:Po,主演:Monkey,主演:Tigress符合电影-主演;类属:大熊猫符合主演-类属;
过滤后的结果Q={主演:Po,电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫}。
S27:判断Q的基数是否大于最小推荐数量阈值,如果大于,则执行S29,如果不大于,则执行S28;
该实施例中在S26中的Q的基数为5个,小于实施例中的最小推荐数量阈值6个,所述“基数”表示为集合中信息数量。执行S28;
S28:扩展次数是否小于最大扩展次数阈值,如果小于,则执行S29;如果不小于,则执行S24;
该实施例中,由于扩展次数为1,小于实施例中的扩展次数2;所以跳转执行S24;
在实施例中,再次执行扩展关联的过程中,需要将候选推荐信息集合中的元素逐个作为所述关键词,执行所述关联扩展步骤,并将产生的关联词添加到所述候选推荐信息集合,根据用户喜好信息对集合中的元素进行过滤。
例如,在实施例中,继续进行关联信息扩展步骤。逐个取Q中的元素进行关联信息扩展,将所有的扩展结果加入Q中。根据图4中的关联关系,候选推荐集合中的各个元素逐个作为关键词,再次关联后得到的候选推荐信息集合。
例如,在图8中,进行二次关联过程中,通过一次关联到的类属:大熊猫,关联到主演:Po、动物园:武汉动物园,动物园:亚特兰大动物园,通过一次关联到的电影:功夫熊猫,关联到上映时间:2008年,制作人:Jonathan Aibel,导演:John Steffensen,译制:上海电影译制厂,主演:Po。
Q={主演:Po,电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫,上映时间:2008年,制作人:Jonathan Aibel,导演:John  Steffensen,译制:上海电影译制厂,分布区域:中国西南地区,外形:丰腴富态,头圆尾短,毛色黑白相间分明,动物园:武汉动物园,动物园:亚特兰大动物园}
根据用户喜好的集合F对候选推荐信息集合Q过滤;
Q=F∩Q
={主演:Po,电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫,动物园:武汉动物园,动物园:亚特兰大动物园}
Q的基数|Q|=7>实施例的推荐数量阈值6;执行S28;
S29:将Q中除用户选择的初始关键词外的词推荐给用户;
Q={主演:Po,电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫,动物园:武汉动物园,动物园:亚特兰大动物园}
所述推荐信息为Q除去初始关键词“主演:Po”,也就是{电影:功夫熊猫,主演:Monkey,主演:Tigress,类属:大熊猫,动物园:武汉动物园,动物园:亚特兰大动物园}。
优选地,上述实施例中关联扩展步骤可以根据需要执行一次或多次,通过设定的关联的次数阈值调节。例如,设定为2次。
另一方面,如果对中心词进行了两次扩展后仍然没有达到推荐数量,则将Q中的除用户选择的初始关键词外的词推荐给用户。这样,由于最大扩展次数的限定,有效的控制了这个推荐流程,而不至于无限的循环下去。
通过上述步骤,得到以下的推荐信息:
当用户查看对象在时刻:00:12:00附近
电影:功夫熊猫时,如果暂停,可以向用户显示:
查看对象信息:
用户选择:主演:Po
扩展1次关联可推荐:
电影:功夫熊猫,
主演:Monkey,
主演:Tigress,
类属:大熊猫
扩展2次关联可推荐:
类属:大熊猫
电影:功夫熊猫,
主演:Monkey,
主演:Tigress,
动物园:武汉动物园
动物园:亚特兰大动物园
图5显示了一些说明性实施例中的信息推荐系统示意图,用户正在收看、收听原始媒体的任何内容的时候,查看到的对象可以是信息描述和文本窗口中的任何信息。
图中的信息描述是一些对用户正在收看、收听的原始媒体的任何内容的一些概括性的描述信息,能够提供持续的信息以供用户选择;例如视频、音频,文本等等,对这些媒体信息,可以有一些基础信息的描述,以电影为例,一般有导演、制片、主要演员、内容简介,相关介绍等等。
图中的文本窗口则是用户能够直接观察到的原始媒体中的信息,即某一时刻的词信息。还是以电影为例,对于任何一个电影场景,都有人物、事件、地点,背景等等,特别是画面或台词中的一些附加信息,例如景点、服饰、引用,以及一些历史、文化等等。这些内容都是随着时间不断变化的,可以作为文本窗口中的内容。也就是当用户暂停时,所关注场景中的该暂停时刻的具体内容。
所述的信息描述和文本窗口是原始媒体的流动信息文本表示,进行流动信息文本表示的主要目的是便于使用原始媒体的用户能够从查看的信息迅速关联至感兴趣的内容,这些感兴趣的内容一般都来自于当前较短的时间窗口中出现的内容。不管是信息描述,还是文本窗口,其中的内容都可以表示为对当前信息所围绕内容的说明,这些说明包括各种组成部分,也可以包括各种关联信息。所有流动信息文本表示由描述对象作为初始,将关联信息逐步扩展,由关联信息组成一个松散的集合,其中数据相关性就是主要通过Content和relation的替代转换进行扩展,也就是说,流动信息文本表示是一些Content组成的集合,而一个Content中的relation又可以是集合中其他元素的Content,通过这种转换关系,可以扩展用户观看部分或者收听部分至各种尽量多的方面,这样,就形成了一个如图4所示的简单 的以关联和关联词为基本组成元素的数据相关网络,而所述数据相关网络正是通过同一数据层层展开而又彼此关联。
用户喜好是用户设定的个人偏好信息,及实施例中提到的集合F存储用户的喜好信息,即用户关注的对象信息的集合F,可以直接提供用户选择,或者从用户查看的信息进行搜集。
采用上述实施例,辅助用户快速获得正在查看对象的感兴趣关联信息推荐,提高准确性,减少打扰,提高效率;当用户通过选择确定了关注的内容点后,能够快速提供关联信息供用户选择,减少用户手动输入的麻烦,提高用户获取感兴趣内容的准确率和效率。
以上所述仅为本发明的优选实施例而已,并不用于限制本发明,对于本领域的技术人员来说,本发明可以有各种更改和变化。凡在本发明的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本发明的保护范围之内。

Claims (7)

  1. 一种向用户推荐信息的方法,其特征在于,包括:
    获取用户查看对象的初始关键词,加入候选推荐信息集合;
    执行以下关联扩展步骤:
    从候选推荐信息集合中逐一获取元素作为关键词,在关联数据库中,查找与所述关键词相匹配的中心词,得到与所述中心词相关联的所有关联词;
    将得到的所有关联词加入候选推荐信息集合,根据用户喜好信息对集合中的元素进行过滤;
    向用户输出候选推荐信息集合中除初始关键词外的所有词。
  2. 根据权利要求1所述的方法,其特征在于,所述获取用户查看对象的初始关键词的过程包括:
    向用户显示查看对象的描述信息和当前文本窗口中的信息,由用户指定一个词作为初始关键词。
  3. 根据权利要求1所述的方法,其特征在于,所述关联数据库包括:
    多个中心词,以及与所述每个中心词相对应的至少一个关联词。
  4. 根据权利要求1所述的方法,其特征在于,所述用户喜好信息可以通过以下方式获得:
    通过软件界面接受用户勾选的用户喜好信息;或者通过记录和分析用户访问的数据,生成用户喜好信息。
  5. 根据权利要求1所述的方法,其特征在于,还包括:如果所述候选推荐信息集合的基数小于数量阈值,则将所述候选推荐信息集合中的元素逐个作为所述关键词,执行所述关联扩展步骤,并将产生的所有关联词添加到所述候选推荐信息集合中,然后根据用户喜好信息对集合中的元素进行过滤。
  6. 根据权利要求5所述的方法,其特征在于,还包括:在判断所述候选推荐信息集合的基数不小于数量阈值或执行所述关联扩展步骤的次数大于次数阈值,则在输出所述候选推荐信息集合中除初始关键词外的所有词之后结束。
  7. 根据权利要求1所述的方法,其特征在于,所述关联扩展的步骤执行一次或多次。
PCT/CN2014/093660 2013-12-31 2014-12-12 向用户推荐信息的方法 WO2015101155A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/779,240 US20160321543A1 (en) 2013-12-31 2014-12-11 Trusted predictive analytic execution middleware

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310754861.4 2013-12-31
CN201310754861.4A CN103761263A (zh) 2013-12-31 2013-12-31 向用户推荐信息的方法

Publications (1)

Publication Number Publication Date
WO2015101155A1 true WO2015101155A1 (zh) 2015-07-09

Family

ID=50528500

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/093660 WO2015101155A1 (zh) 2013-12-31 2014-12-12 向用户推荐信息的方法

Country Status (3)

Country Link
US (1) US20160321543A1 (zh)
CN (1) CN103761263A (zh)
WO (1) WO2015101155A1 (zh)

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103761263A (zh) * 2013-12-31 2014-04-30 武汉传神信息技术有限公司 向用户推荐信息的方法
CN105786467B (zh) * 2014-12-25 2019-08-09 阿里巴巴集团控股有限公司 生成图面的方法及装置
CN106326300A (zh) * 2015-07-02 2017-01-11 富士通株式会社 信息处理方法以及信息处理设备
CN105975471A (zh) * 2015-12-09 2016-09-28 乐视网信息技术(北京)股份有限公司 一种视频推荐方法和系统
CN105468775B (zh) * 2015-12-11 2019-07-23 小米科技有限责任公司 用于电子讲解的方法及装置
CN105760508A (zh) * 2016-02-23 2016-07-13 北京搜狗科技发展有限公司 一种信息推送方法、装置及电子设备
CN106528716A (zh) * 2016-10-26 2017-03-22 腾讯音乐娱乐(深圳)有限公司 一种多媒体搜索内容的推荐方法和装置
CN107229666B (zh) * 2016-12-30 2019-02-26 北京字节跳动科技有限公司 一种基于推荐系统的兴趣探索方法及装置
CN111756754B (zh) * 2017-07-28 2023-04-07 创新先进技术有限公司 一种训练模型的方法及装置
JP7052325B2 (ja) * 2017-12-04 2022-04-12 大日本印刷株式会社 デバイス、セキュアエレメント、プログラム、情報処理システム及び情報処理方法
JP6954084B2 (ja) * 2017-12-18 2021-10-27 大日本印刷株式会社 デバイス、セキュアエレメント、プログラム、情報処理方法及び情報処理システム
CN108153733B (zh) * 2017-12-26 2021-07-09 北京星选科技有限公司 评论质量的分类方法及装置
CN110110207B (zh) * 2018-01-18 2023-11-03 北京搜狗科技发展有限公司 一种信息推荐方法、装置及电子设备
CN110489526A (zh) * 2019-08-13 2019-11-22 上海市儿童医院 一种用于医学检索的检索词扩展方法、装置及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101923545A (zh) * 2009-06-15 2010-12-22 北京百分通联传媒技术有限公司 一种个性化信息推荐的方法
CN102300126A (zh) * 2010-06-24 2011-12-28 日立民用电子株式会社 动态图像推荐系统和动态图像推荐方法
CN103324641A (zh) * 2012-03-23 2013-09-25 日电(中国)有限公司 信息记录推荐方法和装置
CN103761263A (zh) * 2013-12-31 2014-04-30 武汉传神信息技术有限公司 向用户推荐信息的方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090187467A1 (en) * 2008-01-23 2009-07-23 Palo Alto Research Center Incorporated Linguistic extraction of temporal and location information for a recommender system
CN102063433A (zh) * 2009-11-16 2011-05-18 华为技术有限公司 相关项推荐方法和装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101923545A (zh) * 2009-06-15 2010-12-22 北京百分通联传媒技术有限公司 一种个性化信息推荐的方法
CN102300126A (zh) * 2010-06-24 2011-12-28 日立民用电子株式会社 动态图像推荐系统和动态图像推荐方法
CN103324641A (zh) * 2012-03-23 2013-09-25 日电(中国)有限公司 信息记录推荐方法和装置
CN103761263A (zh) * 2013-12-31 2014-04-30 武汉传神信息技术有限公司 向用户推荐信息的方法

Also Published As

Publication number Publication date
US20160321543A1 (en) 2016-11-03
CN103761263A (zh) 2014-04-30

Similar Documents

Publication Publication Date Title
WO2015101155A1 (zh) 向用户推荐信息的方法
US10650861B2 (en) Video summarization and collaboration systems and methods
WO2017096877A1 (zh) 一种推荐方法和装置
US9253511B2 (en) Systems and methods for performing multi-modal video datastream segmentation
US8782056B2 (en) Method and system for facilitating information searching on electronic devices
US20170199943A1 (en) User interface for multivariate searching
US20160014482A1 (en) Systems and Methods for Generating Video Summary Sequences From One or More Video Segments
US20150301718A1 (en) Methods, systems, and media for presenting music items relating to media content
CN108292322B (zh) 使用从查看环境捕捉的信号的媒体数据文件的组织、检索、注释和呈现
JPWO2006019101A1 (ja) コンテンツ関連情報取得装置、コンテンツ関連情報取得方法、およびコンテンツ関連情報取得プログラム
WO2003088665A1 (fr) Dispositif d&#39;edition de metadonnees, dispositif de reproduction de metadonnees, dispositif de distribution de metadonnees, dispositif de recherche de metadonnees, dispositif d&#39;etablissement de conditions de reproduction de metadonnees, et procede de distribution de metadonnees
CN109165302A (zh) 多媒体文件推荐方法及装置
JP2010536233A (ja) マルチメディアコンテンツの推薦キーワード生成システム及びその方法
CN103678668A (zh) 相关搜索结果的提示方法、服务器及系统
JP2006163877A (ja) メタデータ生成装置
US20220107978A1 (en) Method for recommending video content
Lian Innovative Internet video consuming based on media analysis techniques
KR102252522B1 (ko) 내용 기반 동영상 목차 자동생성 방법 및 시스템
WO2014103374A1 (ja) 情報管理装置、サーバ及び制御方法
Redondo-García et al. Augmenting TV newscasts via entity expansion
TWI725375B (zh) 資料搜尋方法及其資料搜尋系統
Bailer et al. Multimedia Analytics Challenges and Opportunities for Creating Interactive Radio Content
Liem et al. MuseSync: standing on the shoulders of Hollywood
Knauf et al. Produce. annotate. archive. repurpose-- accelerating the composition and metadata accumulation of tv content
KR20110071749A (ko) 컨텐츠 정보 관리 장치 및 방법

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: 14875910

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 14779240

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14875910

Country of ref document: EP

Kind code of ref document: A1