CN102982101B - The method of online community users, the context of the body of the push-based services - Google Patents

The method of online community users, the context of the body of the push-based services Download PDF

Info

Publication number
CN102982101B
CN102982101B CN 201210436612 CN201210436612A CN102982101B CN 102982101 B CN102982101 B CN 102982101B CN 201210436612 CN201210436612 CN 201210436612 CN 201210436612 A CN201210436612 A CN 201210436612A CN 102982101 B CN102982101 B CN 102982101B
Authority
CN
Grant status
Grant
Patent type
Prior art keywords
context
user
similarity
tree
attribute
Prior art date
Application number
CN 201210436612
Other languages
Chinese (zh)
Other versions
CN102982101A (en )
Inventor
张晓滨
Original Assignee
西安工程大学
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
Grant date

Links

Abstract

本发明公开了一种基于用户情境本体的网络社区用户推送服务的方法,建立本体化用户情境模型,并对用户综合情境信息进行更新与合成,根据当前用户综合情境的匹配程度向用户进行服务推送;其具体步骤包括:步骤1,本体化用户情境模型的建立;步骤2,用户综合情境信息的更新与合成;步骤3,情境相似性匹配;步骤4,应用服务推送。 The present invention discloses a method for contextual model online community user user context of the body of the push-based services, the establishment of the body of the user, and updates the user with a comprehensive synthesis of contextual information, push services to users based on the degree of matching the current context of a comprehensive user ; specific steps include: establishing step 1, the body of the user context model; step 2. synthesis update user context information integrated; step 3, contextual similarity matching; step 4, a push service application. 本发明通过建立本体化用户综合情境模型,对用户活动有关的特征信息加以描述,可以更好地了解网络社区用户的兴趣,以此为基础,通过用户综合情境的更新与合成,并对情境本体树的相似性进行匹配,实现了将用户应用服务推送给相关用户,能够很好地应用与当前网络社区个性化信息服务中。 The present invention by establishing a comprehensive body of the user context model, feature information related to user activity be described, can better understand the user's online community interest, as a basis, by updating the user with a comprehensive synthesis of context, and the context of the body similarity tree matching, achieved the related user application services pushed to the user, it can be well applied to the current online community personalized information service.

Description

基于用户情境本体的网络社区用户推送服务的方法 The method of online community users, the context of the body of the push-based services

技术领域 FIELD

[0001] 本发明属于互联网社区个性化服务技术领域,具体涉及一种基于用户情境本体的网络社区用户推送服务的方法。 [0001] The present invention belongs to the technical field of personalized Internet community service, particularly to a method based on user context ontology online communities push service.

背景技术 Background technique

[0002] 随着互联网社区网站的飞速发展,通过收集和分析用户的信息来学习用户的兴趣和行为,建立对用户兴趣的描述,研究不同用户的兴趣,主动为用户推荐最需要的资源,从而实现个性化推荐服务是目前互联网社区发展的一个重要方面。 [0002] With the rapid development of the Internet community site, to learn the user's interests and behavior by collecting and analyzing information about the user to establish a description of the user's interests, interests of different users, the initiative to recommend resources most needed for the user, thereby personalized recommendation service is an important aspect of the current Internet community development.

[0003] 目前个性化推荐服务分主要为基于内容推荐、协同过滤推荐、基于知识推荐、基于效用推荐、基于关联规则推荐、混合推荐系统几种方式。 [0003] At present personalized recommendation service points mainly content-based recommendation, collaborative filtering, recommendation based on knowledge, based on utility recommendation, the recommendation based on association rules, hybrid recommendation system in several ways. 其中基于内容的推荐根据项目相关属性特征的定义,实现基于项目属性及其项目之间关联关系和用户的个人喜好的个性化推荐。 Which by definition based on the recommendations of the project-related property characteristics, to achieve personalized recommendations based on the project between project attributes and their relationship and the user's personal preferences. 基于协同过滤推荐根据和自己有着相似爱好的邻居用户所喜欢的项目,自己也同意喜欢的原理来进行推荐,能为用户发现新的感兴趣的项目。 To make a recommendation based on collaborative filtering recommendation based on their own and have similar hobbies neighbor user's favorite project, and he agreed to the principle of love, discover new items of interest to the user. 基于知识推荐是在统一的语义互联环境中获得用户知识和项目知识,通过功能知识的推理或语义匹配项用户推荐。 It is to obtain knowledge of the user and project knowledge in a unified semantic knowledge-based recommendation connected environment, recommended by reasoning or semantic function matches a user's knowledge. 基于效用推荐是建立在对用户使用项目的效用基础上的,其核心问题是怎么为用户创建一个效用函数,然后代入用户和项目等参数根据效用值排序向用户推荐排名靠前的项目。 Recommended based on utility is built on the basis of the use of utility programs on the user, the core issue is how to create a utility function for the user, and then substituted into the parameters of users and projects and other projects to the front of the utility users recommend ranked according to value order. 基于关联规则的推荐技术是以关联规则为基础,把已应用项目作为规则头,推荐项目作为规则体,挖掘数据集中项和项的之间可能存在的相关性,从而将有相关性的项目推荐给用户。 Recommendation technology is based on association rules based, the project has been applied as a rule the head, body recommended the project as a rule, mining correlation may exist between the item and the item data set, which has the relevance of the projects recommended to the user.

[0004] 由于用户的需求受当时用户所处的环境、浏览网页信息、天气变化以及用户情绪的影响,上述几种方法都不能根据网络社区用户兴趣的用户情境,为用户推荐用户所需要的资源,不能很好的应用于当前网络社区个性化信息服务中。 [0004] Due to the needs of users affected by the environment in which the user was browsing the web information, as well as changes in the weather affect the user's mood, neither of these online communities can not be based on user interest in the context of the user, the user is recommended to the resources needed not well apply to the current network of community personalized information service.

发明内容 SUMMARY

[0005] 本发明的目的是提供一种基于用户情境本体的网络社区用户推送服务的方法,解决了现有方法不能根据网络社区用户兴趣的用户情境,为用户推荐用户所需要的资源的问题。 [0005] The present invention is to provide a method for online communities push service user context ontology-based method can not solve the existing network of community based on user interest user context for the user to recommend resources needed by the user of the problem.

[0006] 本发明所采用的技术方案是:基于用户情境本体的网络社区用户推送服务的方法,建立本体化用户情境模型,并对用户综合情境信息进行更新与合成,根据当前用户综合情境的匹配程度向用户进行服务推送;其具体步骤如下: [0006] The technical proposal of the present invention is: A method online communities user context and services based on the push body, the establishment of user context ontology model, and the user to update context information integrated in the synthesis, according to the current user context matching integrated degree of progress push service to a user; specific steps are as follows:

[0007] 步骤1,本体化用户情境模型的建立; [0007] Step 1, the body of the user to establish a context model;

[0008] 步骤2,用户综合情境信息的更新与合成; [0008] Step 2. Synthesis update user context information integrated;

[0009] 步骤3,情境相似性匹配; [0009] Step 3, contextual similarity matching;

[0010] 将源情境本体树和目标本体树在子属性树切分之后,依次对子属性树进行匹配计算,求得源情境本体树根结点与目标情境本体树根结点的相似度; [0010] The context ontology source tree and the target body after the sub-tree segmentation property tree, the property tree matching sub-sequence is calculated, to obtain the source and the target context ontology tree root tree root context similarity body;

[0011] 步骤4,应用服务推送; [0011] Step 4, a push service application;

[0012] 将所有源情境与目标情境的相似度按照从高到低进行排序,将Top-N用户应用的服务推荐给当前用户。 [0012] The similarity of all the source context and the target context sorted by descending, the Top-N user application services recommended to the current user.

[0013] 其中,步骤1中用户情境模型用本体概念树表示,本体概念树中的每一个节点表示了用户情境项的某一个元素; [0013] wherein, in Step 1 is represented by the model user context ontology concept tree, ontology concept each node in the tree represents one item of user context elements;

[0014] 用户情境模型形式化的表达为:UC(ti) = (Dim1 (Attr11,Attr12,Attr13),Dim2,…), [0014] The expression of user context as a formal model: UC (ti) = (Dim1 (Attr11, Attr12, Attr13), Dim2, ...),

[0015] 其中,Ucai)表示用户某时刻^的情境,Dim1表示情境的第i个维度,Attr表示情境维度i的第j个属性,若干个情境信息项和其属性是每个情境维度的构成要素。 [0015] wherein, UCAI) represents a user at a certain time ^ context, denotes the i th dimension Dim1 context, the Attr represents the j-th attribute of the context dimension i, a plurality of information items and context attributes which are configured each context dimension elements.

Figure CN102982101BD00071

个用户情境赋予一个权重,表示用户兴趣变化的逐步遗忘,用来更新用户的情境信息; A user context gives a weighting to indicate the user's interest changes gradually forgotten, used to update the user's context information;

[0017] 其中,f(t)为非线性逐步遗忘函数,表示为: [0017] where, f (t) is a non-linear function gradually forgotten, is expressed as:

[0018] [0018]

Figure CN102982101BD00072

[0019] t:进入当前情境信息的时间,通过用户点击当前浏览网页的绝对时间与参照时间的时间间隔来衡量; [0019] t: time of the current context information entered by the user clicks on the current browsing time and absolute time reference time interval measured;

[0020] max:max(context_time-start_time)最大时间间隔; [0020] max: max (context_time-start_time) the maximum time interval;

[0021] min:min(context_time-start_time)最小时间间隔; [0021] min: min (context_time-start_time) minimum time interval;

[0022] start_time:参照时间,即用户刚开始浏览一系列网页时的最初的开始时间; [0022] start_time: referring to the time that the user just start browsing the first time at the start of a series of web pages;

[0023] context_time:用户进入当前情境的绝对时间用户浏览当前网页点击时间来衡量; [0023] context_time: the user enters the current situation of absolute time users to browse the current webpage Click time to measure;

[0024] tg:指当前浏览网页情境的浏览时间间隔; [0024] tg: refers to the current situation of browsing the web browsing time interval;

[0025] 指浏览商品网页的最小的浏览时间间隔; [0025] refers to browse product pages minimum viewing time interval;

[0026] 指浏览商品网页的最大的浏览时间间隔; [0026] refers to browse product pages maximum browsing time interval;

[0027] G为时间遗忘系数,它反映了f(t)的遗忘能力越大,遗忘越快,反之越慢。 [0027] G is the time forgetting coefficient, which reflects the greater the f (t) is the ability to forget, forgotten faster Instead slower.

[0028] 其中,步骤2中,用户综合情境信息的合成需要考虑到用户的情境属性,情境属性包括标量属性和数量属性,标量属性需计算每一个属性对应的综合情境,合成公式为: [0028] wherein, in step 2, the synthesis requires the user a comprehensive context information taking into account the context of the user attribute, the context attributes include scalar properties and the number of attributes, scalar property to be calculated for each attribute corresponding to the integrated scenario, the synthesis formula is:

Figure CN102982101BD00073

[0030] 数量属性通过计算所有的浏览的网页与相应网页的权量乘积的总和与相应权重的和的比值得到,合成公式为: [0030] The weight and the sum of the respective weights by calculating the ratio of the number of attributes of all web pages viewed in an amount corresponding to the product obtained weights, the formula is synthesized:

Figure CN102982101BD00074

[0032] 其中P1表示第i个浏览网页对应的属性值;f(tJ表示第i个网页对应属性的遗忘权重值。 [0032] where P1 denotes the i th attribute value corresponding to the web browser; f (tJ represented forgetting weighting value corresponding to the i-th page attributes.

[0033]其中,步骤3中对子属性树进行匹配计算步骤如下: [0033] wherein step 3 is calculated to match the property tree sub steps as follows:

[0034] (1)取目标情境和源情境的任一子属性树,其中源情境子属性本体树用Q表示,目标情境子属性本体树用T表示; [0034] (1) take certain context and context source property of any sub-tree, the context in which the source is represented by sub-attributes ontological tree Q, the target sub-attribute context ontology tree represented by T;

[0035](2)获取T中的某一概念Ti,如果其存在转向(3),否则结束; [0035] (2) T acquired in a concept Ti, if present steering (3), else end;

[0036](3)在Q中查找与Ti对应的概念Qi,如果其存在则转向(4),否则转向⑵; [0036] (3) Find the corresponding concept Qi and Ti in Q, if it is present then the steering (4), else go ⑵;

[0037] (4)计算概念Ti和Qi所包含的属性相似度Pi; [0037] (4) computing concepts contained Ti and Qi attribute similarity Pi;

[0038] (5)给⑷计算的每个属性相似度Pi赋予不同的权重Wi ; [0038] (5) for each attribute ⑷ calculated similarity Pi given different weights of the Wi;

[0039] (6)算所有的属性相似度的加权和得到综合相似度Sim(Q,T)=EWi*Pi; [0039] (6) all the calculated weighted similarity of attributes and similarity to comprehensive Sim (Q, T) = EWi * Pi;

[0040] (7)计算库中所有源情境子属性本体树与目标情境子属性本体树的综合相似度Sim(Q,T); [0040] (7) integrated similarity calculated for all the source library context ontology tree sub-attributes to the target sub-attribute context ontology tree Sim (Q, T);

[0041] (8)找到该子属性本体树的父节点概念,再重复(1)~(7)步骤,直至求得最后的源情境本体树根结点与目标情境本体树根结点的相似度。 [0041] (8) sub-attribute body locate the parent node tree concept, repeat (1) to (7) step, until the last source determined context ontology tree root tree root to the target scenario is similar to the body degree.

[0042] 其中,步骤3中源情境本体树和目标本体树在子属性树的切分过程中会形成叶子属性节点树和非叶子属性节点树,叶子属性节点树是由一个节点构成的树,非叶子属性节点树是由多个节点构成的链表树; [0042] wherein, in step 3 the source context ontology tree and the target ontology tree during the cutting process of the sub-attribute tree formed leaf attribute node tree and non-leaf attribute node tree leaf attribute node tree is a tree constituted by one node, non-leaf node of the tree is a linked list attribute tree consisting of a plurality of nodes;

[0043] 叶子属性节点树的相似度是求对应的相同的属性的取值的相似度,即源情境的任一属性节点V和目标情境V'对应属性节点的相似度S(V,V'),根据情境的属性取值类型的不同分为标量属性叶子节点的相似度、数量属性叶子节点的相似度和范围属性叶子节点的相似度; [0043] The similarity of the leaf node of the tree is the attribute values ​​corresponding to the same request attribute similarity, i.e., any source of context attribute nodes V and a target scenario V 'corresponding to the nodes attribute similarity S (V, V' ), depending on the type of the attribute values ​​into the context of similarity, similarity of the number of attributes and similarity of leaf nodes range attribute scalar properties leaf node of the leaf node;

[0044] 非叶子属性节点树的相似度包括名称相似度、属性相似度、实例相似度和结构相似度。 Similarity [0044] non-leaf node of the tree includes a name attribute similarity attribute similarity, structural similarity, and the similarity examples.

[0045] 标量属性叶子节点的相似度的计算公式为: [0045] The scalar properties of the leaf node is calculated similarity:

Figure CN102982101BD00081

[0047] 数量属性叶子节点的相似度的计算公式为: [0047] The number of attributes of leaf nodes similarity is calculated:

Figure CN102982101BD00082

[0049] 范围属性叶子节点的相似度在计算之前,首先对用户的区间[r1,!^]进行规范化 [0049] The similarity range property before computing leaf nodes, the user first interval [r1,! ^] Normalizes

Figure CN102982101BD00083

[0053]其中,Sim名称(X,Y)指概念X和Y的名称相似度,X和y分别表示X和Y具有的同义词集合,IXPlyI指同义词集合X和y的交集的节点个数,Ix_yI指属于集合X但不属于集合y的个数,|yx|指属于集合y但不属于集合x的元素个数,a为比例因子,指集合x和y不相交的元素个数的比例; [0053] wherein, Sim names (X, Y) the name of the concept of similarity of X and Y, X and y represent X and Y have the synsets, IXPlyI refers to the number of nodes synsets of intersection of X and y, Ix_yI refers to belong to the set X, but the number does not belong to the set y, | YX | y refers to, but not belonging to the set number of elements of a set of x, a is a scale factor, the number of elements set refers to the ratio of x and y disjoint;

[0054] 属性相似度的公式为: [0054] attributes similarity formula is:

[0055] Sim属性(Xi,y.) =W1Si(Xi,y.) +w2s2 (Xi,y.) +w3s3 (Xi,y.), [0055] Sim property (Xi, y.) = W1Si (Xi, y.) + W2s2 (Xi, y.) + W3s3 (Xi, y.),

[0056] 其中,xjPy分别指概念X和Y的属性,Sim(Xi,yj指两个属性间的相似度。 [0056] wherein, xjPy properties refer to the concept of X and Y, Sim (Xi, yj refers to the similarity between two attributes.

Figure CN102982101BD00091

[0058] 其中,Sim^ (X,Y)指X,Y两个概念的实例相似度,p(X,Y)指任意一个实例属于X也属于Y的概率,MJT,歹)指属于X但不属于Y的概率,批无,7)指属于Y但不属于X的概率; [0058] wherein, Sim ^ (X, Y) means X, Y two examples of the concept of similarity, p (X, Y) belongs to refer to any instance of X is also a probability of belonging to the Y, MJT, bad) means belongs to X but Y is not a probability, no batch, 7) means the probability of X belonging to Y but does not belong to;

[0059] 结构相似度的公式: [0059] The structural similarity of the formula:

Figure CN102982101BD00092

[0062] Sim结构(X,Y)指X,Y两个概念的结构相似度,Sim祖先节点(X,Y)指X,Y两个概念的祖先节点的相似度,Ancestor(x)指概念节点X的祖先节点的集合,Ancestor(y)指概念节点y的祖先节点的集合;Sirnm^ (X,Y)指X,Y两个概念的子孙节点的相似度,Child(x) 指概念X的子孙节点集合,ChiId(y)指概念y的子孙节点集合。 [0062] Sim structure (X, Y) means X, Y two structural similarity concepts, Sim ancestor node (X, Y) means X, Y similarity ancestor node of two concepts, Ancestor (x) refers to the concept set ancestor node of node X, ancestor (y) refers to the collection of ancestor node of the node y concept; Sirnm ^ (X, Y) means X, the similarity descendant node Y two concepts, Child (x) refers to the concept X the descendants of a node set, ChiId (y) refers to the descendants of the node set concept y.

[0063] 源情境本体树根结点与目标情境本体树根结点的相似度的计算由名称、属性、实例及结构四种相似度的综合计算得到,计算公式为: [0063] calculating the similarity source tree root to the target body the context ontology tree root context is calculated by a combination of four kinds of obtained similarity names, attributes, structures and examples, is calculated as:

[0064] Sim(X,Y) =aSim名称(X,Y) + 0Sim属性(X,Y) +ySim实例(X,Y) + 0Sim结构(X,Y) [0065] 其中,a,P,y,0分别表示从本体概念名称、属性、实例及结构方面的相似度对综合结果产生的影响系数。 [0064] Sim (X, Y) = aSim name (X, Y) + 0Sim property (X, Y) + ySim instance (X, Y) + 0Sim structure (X, Y) [0065] wherein, a, P, y, 0 respectively represent the comprehensive influence coefficient resulting from the similarity of names, attributes, and examples of the concept of a body structure.

[0066] 本发明的有益效果是:本发明通过建立本体化用户综合情境模型,对用户活动有关的特征信息加以描述,可以更好地了解网络社区用户的兴趣,以此为基础,通过用户综合情境的更新与合成,并对情境本体树的相似性进行匹配,实现了将用户应用服务推送给相关用户,能够很好地应用与当前网络社区个性化信息服务中。 [0066] The invention has the advantages that: the invention by establishing a comprehensive body of the user context model, feature information related to user activity be described, can better understand the user's online community interest, on this basis, through an integrated user update and synthesis context, and similarity matching context ontology tree to achieve a user application services pushed to the associated user, it can be well applied to the current online community personalized information service.

附图说明 BRIEF DESCRIPTION

[0067] 图1是本发明情境本体子属性树匹配的流程图。 [0067] FIG. 1 is a flowchart illustrating a context tree matching attribute sub-body of the present invention.

具体实施方式 detailed description

[0068] 下面结合附图和具体实施方式对本发明进行详细说明。 [0068] The present invention will be described in detail in conjunction with accompanying drawings and specific embodiments.

[0069] 一种基于用户情境本体的网络社区用户推送服务的方法,通过建立本体化用户情境模型,并对用户综合情境信息进行更新与合成,根据当前用户综合情境的匹配程度向用户进行服务推送。 [0069] A method of online community users, the context of the body of the push-based services through the establishment of the body of the user context model, and user context information is updated and comprehensive synthesis, a push service to the user according to the degree of matching the current context of a comprehensive user . 用户情境主要是用来描述与用户活动有关的特征信息,它可以是用户基本信息、地点、天气、时间或者与应用程序相关的物理或虚拟的社会、业务等因素,也可以是应用主题用户的心理、个人喜好、情绪等内部的信息。 User context is used to describe the main features of information related to user activity, it can be a basic user information, location, weather, time, or associated with the application of physical or virtual, social, business and other factors, it may also be a subject of the user's application information within the psychological, personal preferences, mood and so on.

[0070] 该方法的具体步骤如下: [0070] In particular steps of the method are as follows:

[0071] 步骤1,本体化用户情境模型的建立; [0071] Step 1, the body of the user to establish a context model;

[0072] 由于用户情境包括不同角度的情境,而每种情境又由不同的属性或者若干的情境要素构成,即情境具有多维度的层次结构,用户情境信息层次结构可对应于本体概念树,本体概念树中的每一个节点表示了用户情境项的某一个元素; [0072] Since the user context comprises a context different angles, and each in turn are composed of different contexts or context attributes of several elements, i.e., hierarchical multi-dimensional context, user context information may correspond to a hierarchical ontology concept tree, ontology each concept tree node represents an element of a certain user context item;

[0073]用户情境模型形式化的表达为:UcUi) = (Dim1 (Attr11,Attr12,Attr13),Dim2, •••), [0073] The expression of user context as a formal model: UcUi) = (Dim1 (Attr11, Attr12, Attr13), Dim2, •••),

[0074] 其中,ucU1)表示用户某时刻^的情境,Dim1表示情境的第i个维度,Attr表示情境维度i的第j个属性,若干个情境信息项和其属性是每个情境维度的构成要素。 [0074] wherein, ucU1) represents a user at a certain time ^ context, denotes the i th dimension Dim1 context, the Attr represents the j-th attribute of the context dimension i, a plurality of information items and context attributes which are configured each context dimension elements.

[0075] 步骤2,用户综合情境信息的更新与合成; [0075] Step 2. Synthesis update user context information integrated;

[0076] 用户的行为具有一定的连续性,而用户当前的情境信息仅仅反映了用户短时间内的信息需求,随着时间的不断积累用户的情境信息也不断的积累变化形成用户的历史情境uc(t....),即当前用户的综合情境是由历史情境和当前情境两部分构成的。 [0076] the user's behavior with a certain continuity, and the user's current context information only reflects the information needs of users in a short time, the historical context uc With the accumulation time of the user's context information constantly changes form the accumulation of users (t ....), that is, the current overall situation of the user by the historical context and the current situation of two parts. 历史情境是随着时间的积累形成的,为以后的情境信息提供了基础,而当前情境则是当前获取的用户情境信息。 Historical context is formed with the accumulation of time, it provided the foundation for contextual information, and the current situation is the user context information currently acquired.

Figure CN102982101BD00101

予一个权重,表示用户兴趣变化的逐步遗忘,用来更新用户的情境信息; To a weight, representing the user's interests change gradually forgotten, used to update the user's context information;

[0078]f(t)为非线性逐步遗忘函数,表示为: [0078] f (t) is a non-linear function gradually forgotten, is expressed as:

Figure CN102982101BD00102

[0080] t:进入当前情境信息的时间,通过用户点击当前浏览网页的绝对时间与参照时间的时间间隔来衡量; [0080] t: time of the current context information entered by the user clicks on the current browsing time and absolute time reference time interval measured;

[0081] max:max(context_time-start_time)最大时间间隔; [0081] max: max (context_time-start_time) the maximum time interval;

[0082] min:min(context_time-start_time)最小时间间隔; [0082] min: min (context_time-start_time) minimum time interval;

[0083] start_time:参照时间,即用户刚开始浏览一系列网页时的最初的开始时间; [0083] start_time: referring to the time that the user just start browsing the first time at the start of a series of web pages;

[0084] context_time:用户进入当前情境的绝对时间用户浏览当前网页点击时间来衡量; [0084] context_time: the user enters the current situation of absolute time users to browse the current webpage Click time to measure;

[0085] tg:指当前浏览网页情境的浏览时间间隔; [0085] tg: refers to the current situation of browsing the web browsing time interval;

[0086] 指浏览商品网页的最小的浏览时间间隔; [0086] refers to browse product pages minimum viewing time interval;

[0087] t_sas:指浏览商品网页的最大的浏览时间间隔; [0087] t_sas: browse product pages refers to the maximum browsing time interval;

[0088]G为时间遗忘系数,它反映了f(t)的遗忘能力越大,遗忘越快,反之越慢; [0089]用户综合情境信息的合成需要考虑到用户的情境属性,情境属性包括标量属性和数量属性; [0088] G is the time forgetting coefficient, which reflects the greater the f (t) is the ability to forget, forgotten faster Instead slower; synthesis requires [0089] The user context information integrated into consideration user context attributes, including attributes context scalar property and the number of attributes;

[0090] (1)标量属性:需计算每一个属性对应的综合情境,合成公式为: [0090] (1) scalar property: a property to be calculated for each context corresponding to the integrated synthesis formula is:

Figure CN102982101BD00111

[0092] (2)数量属性:通过计算所有的浏览的网页与相应网页的权量乘积的总和与相应权重的和的比值得到,合成公式为: [0092] (2) the number of attributes: weight and the sum of the respective weights by calculating the ratio of all web pages viewed in an amount corresponding to the product obtained weights, the formula is synthesized:

Figure CN102982101BD00112

[0094] 其中P1表示第i个浏览网页对应的属性值;f(tJ表示第i个网页对应属性的遗忘权重值。 [0094] where P1 denotes the i th attribute value corresponding to the web browser; f (tJ represented forgetting weighting value corresponding to the i-th page attributes.

[0095] 步骤3,情境相似性匹配;将源情境本体树和目标本体树在子属性树切分之后,依次对子属性树进行匹配计算,求得源情境本体树根结点与目标情境本体树根结点的相似度。 [0095] Step 3, the affinity matching context; the context ontology source tree and the target body after the sub-tree segmentation property tree, the property tree matching sub-sequence is calculated, to obtain the source and the target context ontology tree root context body tree root similarity.

[0096] 对子属性树进行匹配计算步骤如下,如图1所示: Matching calculation step [0096] the following sub-attribute tree, shown in Figure 1:

[0097] (1)取目标情境和源情境的任一子属性树,其中源情境子属性本体树用Q表示,目标情境子属性本体树用T表示; [0097] (1) take certain context and context source property of any sub-tree, the context in which the source is represented by sub-attributes ontological tree Q, the target sub-attribute context ontology tree represented by T;

[0098](2)获取T中的某一概念Ti,如果其存在转向(3),否则结束; [0098] (2) T acquired in a concept Ti, if present steering (3), else end;

[0099](3)在Q中查找与Ti对应的概念Qi,如果其存在则转向(4),否则转向⑵; [0099] (3) Find the corresponding concept Qi and Ti in Q, if it is present then the steering (4), else go ⑵;

[0100] (4)计算概念Ti和Qi所包含的属性相似度Pi; [0100] (4) computing concepts contained Ti and Qi attribute similarity Pi;

[0101] (5)给⑷计算的每个属性相似度Pi赋予不同的权重Wi; [0101] (5) for each attribute ⑷ calculated similarity Pi given different weights of the Wi;

[0102] (6)算所有的属性相似度的加权和得到综合相似度Sim(Q,T) =EWi*Pi; [0102] (6) all the calculated weighted similarity of attributes and similarity to comprehensive Sim (Q, T) = EWi * Pi;

[0103] (7)计算库中所有源情境子属性本体树与目标情境子属性本体树的综合相似度Sim(Q,T); [0103] (7) integrated similarity calculated for all the source library context ontology tree sub-attributes to the target sub-attribute context ontology tree Sim (Q, T);

[0104] (8)找到该子属性本体树的父节点概念,再重复⑴~(7)步骤,直至求得最后的源情境本体树根结点与目标情境本体树根结点的相似度。 [0104] (8) sub-attribute body locate the parent node tree concept, repeat ⑴ ~ (7) step, until the last source context ontology tree root to the target root context ontology tree similarity obtained.

[0105] 源情境本体树和目标本体树在子属性树的切分过程中会形成叶子属性节点树和非叶子属性节点树,叶子属性节点树是由一个节点构成的树,非叶子属性节点树是由多个节点构成的链表树; [0105] Source context ontology tree and the target ontology tree during the cutting process of the sub-attribute tree formed leaf attribute node tree and non-leaf attribute node tree leaf attribute node tree is a tree, non-leaf attribute node tree consisting of a node tree list is composed of a plurality of nodes;

[0106] 叶子属性节点树的相似度是求对应的相同的属性的取值的相似度,即源情境的任一属性节点V和目标情境V'对应属性节点的相似度S(V,V'),根据情境的属性取值类型的不同分为标量属性叶子节点的相似度、数量属性叶子节点的相似度和范围属性叶子节点的相似度。 [0106] Similarity leaf node of the tree is the attribute values ​​corresponding to the same request attribute similarity, i.e., any source of context attribute nodes V and a target scenario V 'corresponding to the nodes attribute similarity S (V, V' ), depending on the type of property values ​​into context similarity similarity, and the similarity range attribute scalar properties leaf node of the leaf node of the leaf node number of attributes.

[0107] (1)标量属性叶子节点的相似度的计算公式为: [0107] similarity calculation formula (1) scalar property leaf node is:

Figure CN102982101BD00121

[0111] (3)范围属性叶子节点的相似度在计算之前,首先对用户的区间[r1,rn]进行规范 [0111] similarity (3) the range of properties of the leaf node prior to calculating the first interval of the user [r1, rn] regulate

Figure CN102982101BD00122

[0113] 非叶子属性节点树的相似度包括名称相似度、属性相似度、实例相似度和结构相似度。 Similarity [0113] non-leaf node of the tree includes a name attribute similarity attribute similarity, structural similarity, and the similarity examples.

[0114] (1)名称相似度的公式为: [0114] (1) similarity of formula name:

[0115] [0115]

Figure CN102982101BD00123

[0116]其中,Sim名称(X,Y)指概念X和Y的名称相似度,X和y分别表示X和Y具有的同义词集合,IXPlyI指同义词集合X和y的交集的节点个数,Ix_yI指属于集合X但不属于集合y的个数,|yx|指属于集合y但不属于集合X的元素个数,a为比例因子,指集合X 和y不相交的元素个数的比例。 [0116] wherein, Sim names (X, Y) the name of the concept of similarity of X and Y, X and y represent X and Y have the synsets, IXPlyI refers to the number of nodes synsets of intersection of X and y, Ix_yI refers to belong to the set X, but the number does not belong to the set y, | YX | y refers to, but not belonging to the set number of elements of the set X, a is a scale factor means the ratio of the number of elements X and y are set disjoint.

[0117] (2)属性相似度的公式为: [0117] (2) attribute similarity formula is:

[0118]Sim属性(Xi,yj) =W1Si(Xi,yj) +W2S2 (Xi,yj) +W3S3 (Xi,yj), [0118] Sim property (Xi, yj) = W1Si (Xi, yj) + W2S2 (Xi, yj) + W3S3 (Xi, yj),

[0119]其中,xjPy。 [0119] wherein, xjPy. 分别指概念X和Y的属性,Sim属性(Xi,yj指两个属性间的相似度。 Refer to the concept of properties X and Y, Sim properties (Xi, yj refers to the similarity between two attributes.

Figure CN102982101BD00124

[0122] 其中,Sim$_ (X,Y)指X,Y两个概念的实例相似度,p(X,Y)指任意一个实例属于X也属于Y的概率,梦(U)指属于X但不属于Y的概率,指属于Y但不属于X的概率。 [0122] wherein, Sim $ _ (X, Y) means X, Y two examples of the concept of similarity, P (X, Y) belongs to refer to any instance of X is also a probability of belonging to Y, the dream (U) belonging to the means X but the probability does not belong to the Y, the probability that X belongs to Y, but not part of.

[0123] (4)结构相似度的公式: [0123] (4) structural similarity of the formula:

[0124]Sim结构(X,Y) =Sim祖先节点(X,Y)XSim子孙节点(X,Y), [0124] Sim structure (X, Y) = Sim ancestor node (X, Y) XSim descendant nodes (X, Y),

Figure CN102982101BD00131

[0126]Sim结构(X,Y)指X,Y两个概念的结构相似度,Sim祖先节点(X,Y)指X,Y两个概念的祖先节点的相似度,Ancestor(x)指概念节点X的祖先节点的集合,Ancestor(y)指概念节点y的祖先节点的集合;Sirnm^ (X,Y)指X,Y两个概念的子孙节点的相似度,Child(x) 指概念X的子孙节点集合,ChiId(y)指概念y的子孙节点集合。 [0126] Sim structure (X, Y) means X, Y two structural similarity concepts, Sim ancestor node (X, Y) means X, Y similarity ancestor node of two concepts, Ancestor (x) refers to the concept set ancestor node of node X, ancestor (y) refers to the collection of ancestor node of the node y concept; Sirnm ^ (X, Y) means X, the similarity descendant node Y two concepts, Child (x) refers to the concept X the descendants of a node set, ChiId (y) refers to the descendants of the node set concept y.

[0127] 源情境本体树根结点与目标情境本体树根结点的相似度的计算由名称、属性、实例及结构四种相似度的综合计算得到,计算公式为: [0127] calculating the similarity source context tree root to the target body the context ontology tree root node obtained by a combination of four kinds of similarity calculated names, attributes, structures and examples, is calculated as:

[0128]Sim(X,Y) =aSim名称(X,Y) + 0Sim属性(X,Y) +ySim实例(X,Y) + 0Sim结构(X,Y), [0128] Sim (X, Y) = aSim name (X, Y) + 0Sim property (X, Y) + ySim instance (X, Y) + 0Sim structure (X, Y),

[0129] 其中,a,P,Y,0分别表示从本体概念名称、属性、实例及结构方面的相似度对综合结果产生的影响系数。 [0129] wherein, a, P, Y, 0 represent the comprehensive influence coefficient resulting from the similarity of names, attributes, and examples of the concept of a body structure.

[0130] 步骤4,应用服务推送; [0130] Step 4, a push service application;

[0131] 将所有源情境与目标情境的相似度按照从高到低进行排序,将Top-N用户应用的服务推荐给当前用户。 [0131] The similarity of all the source context and the target context sorted by descending, the Top-N user application services recommended to the current user.

Claims (4)

  1. 1.基于用户情境本体的网络社区用户推送服务的方法,其特征在于,建立本体化用户情境模型,并对用户综合情境信息进行更新与合成,根据当前用户综合情境的匹配程度向用户进行服务推送;其具体步骤如下: 步骤1,本体化用户情境模型的建立;用户情境模型用本体概念树表示,本体概念树中的每一个节点表示了用户情境项的某一个元素; 用户情境模型形式化的表示为: Uc (tj = (Dim1 (Attr11, Attr12, Attr13), Dim2,...), 其中,uc U1)表示用户某时刻h的情境,Dim i表示情境的第i个维度,Attr ^表示情境维度i的第j个属性,若干个情境信息项和其属性是每个情境维度的构成要素; 步骤2,用户综合情境信息的更新与合成; 步骤3,情境相似性匹配;将源情境本体树和目标本体树在子属性树切分之后,依次对子属性树进行匹配计算,求得源情境本体树根结点与目标情境 1. Method online community user user context ontology-based push service, characterized in that the establishment of the body of the user context model, and update the user with a comprehensive synthesis of contextual information, push services to users based on the degree of matching the current context of a comprehensive user ; the following steps: establishing step 1, the body of the user context model; user context ontology concept model represented by a tree, the tree ontology concept each node represents an element of a user context item; user context model formal expressed as: Uc (tj = (Dim1 (Attr11, Attr12, Attr13), Dim2, ...), wherein, uc U1) indicating that the user of a time h context, Dim i denotes the i th dimension context, Attr ^ represents j-i of the attribute context dimension, a number of information items and context attributes which are components of each context dimension; step 2. synthesis update user context information integrated; step 3, the affinity matching context; the context ontology source tree and the target body after the sub-tree segmentation property tree, the property tree matching sub-sequence is calculated, to obtain the source and the target root context tree the context ontology 本体树根结点的相似度; 源情境本体树和目标本体树在子属性树的切分过程中会形成叶子属性节点树和非叶子属性节点树,叶子属性节点树是由一个节点构成的树,非叶子属性节点树是由多个节点构成的链表树; 所述叶子属性节点树的相似度是求对应的相同的属性的取值的相似度,即源情境的任一属性节点V和目标情境V'对应属性节点的相似度S (V,V ),根据情境的属性取值类型的不同分为标量属性叶子节点的相似度、数量属性叶子节点的相似度和范围属性叶子节点的相似度; 所述非叶子属性节点树的相似度包括名称相似度、属性相似度、实例相似度和结构相似度; 所述标量属性叶子节点的相似度的计算公式为: Similarity ontology tree root; source tree and the target context ontology tree body during the cutting process of the sub-attribute tree leaf formation properties and non-leaf node of the tree node tree attributes, attribute node tree is a tree leaf constituted by a node , non-leaf node of the tree is a linked list attribute tree composed of a plurality of nodes; the similarity properties of the leaf node of the tree is the similarity of the same attributes corresponding demand values, i.e., any source of a context attribute and the target node V context V 'corresponding to the nodes attribute similarity S (V, V), depending on the type of property values ​​into context similarity similarity, and the similarity range attribute scalar properties leaf node of the leaf node of the leaf node number of attributes ; similarity of the non-leaf node of the tree includes a name attribute similarity attribute similarity, structural similarity, and the similarity example; the similarity is calculated scalar properties leaf node is:
    Figure CN102982101BC00021
    所述数量属性叶子节点的相似度的计算公式为: The number of leaf nodes attribute similarity is calculated:
    Figure CN102982101BC00022
    所述范围属性叶子节点的相似度在计算之前,首先对用户的区间[Ρ,Γη]进行规范化得到.所述范围属性叶子节点的相似度的计算公式为: 7 The range attribute leaf nodes prior to calculating the degree of similarity, the user first interval [Ρ, Γη] to obtain the standardized range attribute leaf nodes similarity is calculated: 7
    Figure CN102982101BC00023
    J J
    Figure CN102982101BC00024
    其中,Sirn^ (X,Υ)指概念X和Y的名称相似度,X和y分别表示X和Y具有的同义词集合,IX Π y I指同义词集合X和y的交集的节点个数,I xy I指属于集合X但不属于集合y的个数,|yx|指属于集合y但不属于集合X的元素个数,α为比例因子,指集合X和y 不相交的元素个数的比例; 所述属性相似度的公式为: Wherein, Sirn ^ (X, Υ) the name of the concept of similarity of X and Y, X and y represent X and Y have the synsets, IX Π y I refers to the number of nodes synsets of intersection of X and y, I xy I belong to the set X refers to the number but does not belong to the set of y, | YX | y belonging to the set, but does not refer to the number of elements of the set X, [alpha] is the scale factor, means the ratio of the number of elements of the set X and y are disjoint ; the attribute similarity formula is:
    Figure CN102982101BC00031
    其中,xjPy_j分别指概念X和Y的属性,Sim嫩(XiJj)指两个属性间的相似度;W^w 2, W3分别指对应于si,s2, S3即属性名称、数据类型和实例的权重,且 Wherein, xjPy_j properties refer to the concept of X and Y, Sim tender (XiJj) refers to the similarity between two attributes; W ^ w 2, W3 denote corresponding to the si, s2, S3 i.e. attribute name, data type, and examples of weight, and
    Figure CN102982101BC00032
    所述实例相似度的公式为: Examples of similarity of the formula:
    Figure CN102982101BC00033
    其中,Sim$_ (X,Y)指X,Y两个概念的实例相似度,p (X,Y)指任意一个实例属于X也属于Y的概率,指属于X但不属于γ的概率,MX, 指属于γ但不属于X的概率; 所述结构相似度的公式: Wherein, Sim $ _ (X, Y) means X, Y two examples of the concept of similarity, P (X, Y) belongs to refer to any instance of X is also a probability of belonging to the Y, but the probability of X belonging to γ ​​means does not belong, MX, the probability that γ but not belonging to X; and the structural similarity of the formula:
    Figure CN102982101BC00034
    Sim结构(X,Y)指X,Y两个概念的结构相似度,Sim祖先节点(X,Y)指X,Y两个概念的祖先节点的相似度,Ancestor(x)指概念节点X的祖先节点的集合,Ancestor (y)指概念节点y 的祖先节点的集合;(X,Y)指X,Y两个概念的子孙节点的相似度,Child(x)指概念X的子孙节点集合,Child(y)指概念y的子孙节点集合; 所述源情境本体树根结点与目标情境本体树根结点的相似度的计算由名称、属性、实例及结构四种相似度的综合计算得到,计算公式为: Sim structure (X, Y) means X, Y two structural similarity concepts, Sim ancestor node (X, Y) means X, Y similarity ancestor node of two concepts, Ancestor (x) refers to the concept node X ancestor node of the set, the ancestor (y) refers to the collection of ancestor node of the node y concept; (X, Y) means X, the similarity descendant node Y two concepts, Child (x) means the set of descendants of the node X of the concept, Child (y) refers to the descendants of the node y concept set; calculating a degree of similarity of the source and the target context ontology tree root tree root context body obtained by a combination of four kinds of similarity calculated names, attributes, structures and examples , calculated as follows:
    Figure CN102982101BC00035
    其中,α,β,γ,Θ分别表不从本体概念名称、属性、实例及结构方面的相似度对综合结果产生的影响系数; 步骤4,应用服务推送;将所有源情境与目标情境的相似度按照从高到低进行排序,将Top-N用户应用的服务推荐给当前用户。 Wherein, α, β, γ, Θ table are not comprehensive influence coefficient resulting from the similarity of names, attributes, and examples of the structure of the ontology concept; step 4, a push service application; all the source and the target context similar situations degree sorted by descending, the Top-N user application services recommended to the current user.
  2. 2.根据权利要求1所述的基于用户情境本体的网络社区用户推送服务的方法,其特征在于,步骤2中用户综合情境信息表示为: 2. The method of online communities user context body of the push service based on claim 1, wherein, in step 2 the user context information integrated expressed as:
    Figure CN102982101BC00036
    其中f(t)给每一个用户情境赋予一个权重,表示用户兴趣变化的逐步遗忘,用来更新用户的情境信息; 所述f(t)为非线性逐步遗忘函数,表示为: Where f (t) to a user context for each given a weight, representing the user's interest changes gradually forgotten, used to update the user's context information; of the f (t) is a non-linear function gradually forgotten, is expressed as:
    Figure CN102982101BC00041
    t:进入当前情境信息的时间,通过用户点击当前浏览网页的绝对时间与参照时间的时间间隔来衡量; max:max(context_time-start_time)最大时间间隔; min:min(context_time-start_time)最小时间间隔; Start_time :参照时间,即用户刚开始浏览一系列网页时的最初的开始时间; contextjime :用户进入当前情境的绝对时间用户浏览当前网页点击时间来衡量; :指当前浏览网页情境的浏览时间间隔; t_砸:指浏览商品网页的最小的浏览时间间隔; 指浏览商品网页的最大的浏览时间间隔; ζ为时间遗忘系数,它反映了f(t)的遗忘能力,ζ越大,遗忘越快,反之越慢。 t: enter the time current context information through a user clicks on the current browsing time absolute time reference time interval measured; max: max (context_time-start_time) the maximum time interval; min: min (context_time-start_time) minimum time interval ; Start_time: referring to the time that the user just start browsing the first time at the start of a series of pages; contextjime: the user enters the current situation of absolute time users to browse the current webpage click time to measure;: refers to the current situation of browsing the web browsing time interval; t_ hit: browse product pages refers to the minimum viewing time interval; refers to browse product pages maximum browsing time interval; ζ is time forgetting coefficient, which reflects the f (t) is the ability to forget, ζ greater forgotten sooner Instead slower.
  3. 3. 根据权利要求2所述的基于用户情境本体的网络社区用户推送服务的方法,其特征在于,步骤2中,用户综合情境信息的合成需要考虑到用户的情境属性,情境属性包括标量属性和数量属性; 所述标量属性需计算每一个属性对应的综合情境,合成公式为: 3. The method of online communities based on user context ontology push service according to claim 2, wherein, in step 2, the synthesis requires consideration user context information integrated into the user context attributes, context attributes include scalar property and the number of attributes; the scalar attribute to be calculated for each context attribute corresponding to the integrated synthesis formula is:
    Figure CN102982101BC00042
    所述数量属性通过计算所有的浏览的网页与相应网页的权量乘积的总和与相应权重的和的比值得到,合成公式为: The number of attributes by weight and the sum of the respective weights of all the calculated ratio with a corresponding web page to browse the right amount of product is obtained, the synthesis of the formula:
    Figure CN102982101BC00043
    其中P1表示第i个浏览网页对应的属性值;f(t J表示第i个网页对应属性的遗忘权重值。 Wherein P1 represents an i-th attribute value corresponding to the web browser; f (t J represents a forgetting the i-th weight value of the corresponding attribute page.
  4. 4. 根据权利要求3所述的基于用户情境本体的网络社区用户推送服务的方法,其特征在于,步骤3中对子属性树进行匹配计算步骤如下: (1) 取目标情境和源情境的任一子属性树,其中源情境子属性本体树用Q表示,目标情境子属性本体树用T表不; (2) 获取T中的某一概念Ti,如果其存在转向(3),否则结束; (3) 在Q中查找与Ti对应的概念Qi,如果其存在则转向(4),否则转向(2); (4) 计算概念Ti和Qi所包含的属性相似度Pi ; (5) 给(4)计算的每个属性相似度Pi赋予不同的权重Wi ; (6) 算所有的属性相似度的加权和得到综合相似度Sim(Q,T) =Σ Wi*Pi ; (7) 计算库中所有源情境子属性本体树与目标情境子属性本体树的综合相似度Sim(Q,T); (8)找到该子属性本体树的父节点概念,再重复(1)~(7)步骤,直至求得最后的源情境本体树根结点与目标情境本体树根结点的相似度。 4. The method of online communities user context and services based on the push body, wherein according to claim 3, the step 3 is calculated to match the property tree sub steps as follows: (1) takes any one of the source and the target context context a sub-attribute tree, the context in which the source is represented by sub-attributes ontological tree Q, the target context tree body with sub-attributes table T is not; (2) T acquires a concept of Ti, if present steering (3), else end; (3) Find Q concepts corresponding to Qi and Ti, if it is present then the steering (4), else go to (2); (4) computing concepts contained Ti and Qi attributes similarity Pi; (5) to ( 4) the degree of similarity calculated for each attribute Pi given different weights Wi; (6) all the calculated weighted similarity of the properties and obtained integrated similarity Sim (Q, T) = Σ Wi * Pi; (7) computing Base comprehensive similarity all source sub-attributes context tree body and the target body sub-attribute context tree Sim (Q, T); (8) body to locate the sub-attribute of the parent node tree concept, repeat (1) to (7) step, until finally determined the source of the context and the target context ontology tree root tree root body of similarity.
CN 201210436612 2012-11-05 2012-11-05 The method of online community users, the context of the body of the push-based services CN102982101B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201210436612 CN102982101B (en) 2012-11-05 2012-11-05 The method of online community users, the context of the body of the push-based services

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201210436612 CN102982101B (en) 2012-11-05 2012-11-05 The method of online community users, the context of the body of the push-based services

Publications (2)

Publication Number Publication Date
CN102982101A true CN102982101A (en) 2013-03-20
CN102982101B true CN102982101B (en) 2015-11-18

Family

ID=47856119

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201210436612 CN102982101B (en) 2012-11-05 2012-11-05 The method of online community users, the context of the body of the push-based services

Country Status (1)

Country Link
CN (1) CN102982101B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105718579A (en) * 2016-01-22 2016-06-29 浙江大学 Information push method based on internet-surfing log mining and user activity recognition

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102208086A (en) * 2010-03-31 2011-10-05 北京邮电大学 Field-oriented personalized intelligent recommendation system and implementation method
CN102236646A (en) * 2010-04-20 2011-11-09 得利在线信息技术(北京)有限公司 Personalized item-level vertical pagerank algorithm iRank

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9002924B2 (en) * 2010-06-17 2015-04-07 Microsoft Technology Licensing, Llc Contextual based information aggregation system

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102208086A (en) * 2010-03-31 2011-10-05 北京邮电大学 Field-oriented personalized intelligent recommendation system and implementation method
CN102236646A (en) * 2010-04-20 2011-11-09 得利在线信息技术(北京)有限公司 Personalized item-level vertical pagerank algorithm iRank

Also Published As

Publication number Publication date Type
CN102982101A (en) 2013-03-20 application

Similar Documents

Publication Publication Date Title
Paliwal et al. Semantics-based automated service discovery
Zhang et al. Collaborative knowledge base embedding for recommender systems
Wang et al. ReCoM: reinforcement clustering of multi-type interrelated data objects
Ristoski et al. Semantic Web in data mining and knowledge discovery: A comprehensive survey
US20110264649A1 (en) Adaptive Knowledge Platform
US20100235307A1 (en) Method, system, and computer program for user-driven dynamic generation of semantic networks and media synthesis
Serdyukov et al. Modeling multi-step relevance propagation for expert finding
Yin et al. Building taxonomy of web search intents for name entity queries
White et al. Predicting user interests from contextual information
De Meo et al. Recommendation of similar users, resources and social networks in a Social Internetworking Scenario
Wang et al. Ranking user's relevance to a topic through link analysis on web logs
Tseng et al. Efficient algorithms for mining top-k high utility itemsets
CN102831234A (en) Personalized news recommendation device and method based on news content and theme feature
Mei et al. A general optimization framework for smoothing language models on graph structures
CN102208086A (en) Field-oriented personalized intelligent recommendation system and implementation method
Delbru et al. Hierarchical link analysis for ranking web data
Zhang GroRec: a group-centric intelligent recommender system integrating social, mobile and big data technologies
CN103136337A (en) Distributed knowledge data mining device and mining method used for complex network
CN103731738A (en) Video recommendation method and device based on user group behavioral analysis
Xie et al. Community-aware user profile enrichment in folksonomy
US20120254096A1 (en) Portable Inferred Interest and Expertise Profiles
Maleszka et al. A method for collaborative recommendation using knowledge integration tools and hierarchical structure of user profiles
Tran et al. SemSearchPro–Using semantics throughout the search process
Ristoski et al. Mining the web of linked data with rapidminer
Salin et al. Using semantic information for web usage mining based recommendation

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model
CF01