Browse Prior Art Database

Method for personalized micro-blog ranking

IP.com Disclosure Number: IPCOM000236900D
Publication Date: 2014-May-21
Document File: 4 page(s) / 114K

Publishing Venue

The IP.com Prior Art Database

Abstract

The mirco-blog is massive and overwhelming in a stream manner. Furthermore, the data is ranked by the timestamp/alphabetic/follower, not personalized or contextualized. The users are prone to miss important or interested content and can not trace it back conveniently. The invention provides a novel method & apparatus to reorganizing & ranking stream micro-blogs for personal users by combining the historical and contexutalized information.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 70% of the total text.

Page 01 of 4

Method for personalized micro

Method for personalized micro-

Micro-blog data is massive and overwhelming in a stream manner, usually the data is ranked by the timestamp, not personalized. Thus it is easy to miss important or interested content and can never trace it back conveniently. This invention provides a novel method & apparatus to reorganizing & ranking stream micro-blogs for personal users.

We propose a user browsing behavior driven learning to rank model to extract user's preference over the streaming micro-blogs, and rank & reorganize micro-blogs. It is based on the observation that user browsing behaviors are informative to extract user's preference.

In particular, the proposed disclosure has the following key components and features:

1) Method for generating labeling to train a personalized scoring model using user's browsing behaviors including valid browsing time over a micro-blog, forward and comment, including

- valid browsing behavior can be determined by cross-checking the information collected from the mobile device system including if the screen is locked, if - the microblog application is switched to background, the location and timing of the browsing

- context sensitive models are separately trained using the location information from other sources like gps, call detail records, etc.

- the order between two samples for training is determined by the frequency of the way that the user browse the samples , instead of pre-defined.

- the browsin...