Browse Prior Art Database

Method of Automatically Quantifying and Qualifying Social Connections based on the Social Topology, Activities and Attributes of Each Social Node

IP.com Disclosure Number: IPCOM000200639D
Publication Date: 2010-Oct-22
Document File: 3 page(s) / 60K

Publishing Venue

The IP.com Prior Art Database

Related People

Bo Long: INVENTOR [+2]

Abstract

A method of automatically quantifying and qualifying social connections based on the social topology, activities and attributes of each social node is disclosed.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Method of Automatically Quantifying and Qualifying Social Connections based on the Social Topology, Activities and Attributes of Each Social Node

Abstract

A method of automatically quantifying and qualifying social connections based on the social topology, activities and attributes of each social node is disclosed.

Description

Disclosed is a method of automatically quantifying and qualifying social connections based on the social topology, activities and attributes of each social node.  The method involves creation of a Link Profile (LP).  The LP is a concise and accurate representation of a link between two users, which describes the type(s) associated the link and strength associated with the link.  Thereafter, a Mixed Membership Social Targeting (MMST) model is created.  The MMST represents a general probabilistic model and is illustrated in the figure.

Figure

Here, Xi(uf) and  Xi’(uf)  denote user features (attribute and activity information) for user i and i’;

Li(k) is the latent variable that denotes membership of user i within  kth type of user community .  Here totally there are k types of communities.  For example, first type is an alumni community and second type is a professional community, etc.

Ω(u) denotes the prior distribution on latentvariables,Li(k)

Xii’(uu)  denotes the link between two users i and i’, which depends on the latent variables,   Li(i)  …. Li(k)   and Li’(i)  …. Li’(k) , i.e., the link depends on what types of communities users belong to and how strongly users belong to the communities.  Hence, Xii’(uu)   is  the  link profile which provides the types of links and strength level associated with the links.

In order to recommend items to users, items may be included into the model.  For example, Xj(if) and  Xj’(if)  denote item features (attribute and activity information) for item j and j’.  Similarly, Zj(i) and Zj’(i)  denote latent variablesfor item j and j’.  Ω(i) denotes the prior distribution on latent   variables, Zj(i).

Further, Xij(ui)  denotes  the preference of user i on item j;Xi’j’(ui)   denotes the preference of user i’ on item j’.  Hence, when targeting users for item j, through the latent variables   Li(i)  …. Li(k)   and Li’(i)  …. Li’(k) ,  the link profile information is incorporated into the user’s preference on items to achieve more accurate targeting.

In the above model,Xii’(uu) quantifies and qualifies the link between two users.  Therefore, Xii’(uu)  provides information corresponding to the different types of links and strength level associated with the lin...