Browse Prior Art Database

Context adaptive generation of user desired quick links for continuous content

IP.com Disclosure Number: IPCOM000248956D
Publication Date: 2017-Jan-24
Document File: 4 page(s) / 58K

Publishing Venue

The IP.com Prior Art Database

Abstract

The disclosure discloses generate quick links automatically by summarizing content according to the stop time, it can help create quick links for continuous content that are mostly desired by users without any human interference, after determining the sections of interests, these sections could be used to better improve content, or extracting keywords for the categoires of content.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 48% of the total text.

1

Context adaptive generation of user desired quick links for continuous content

In online shopping scenarios, there are often many content in the detailed descriptions of products. These content may not be well structured, and it is hard for shoppers to quickly find key information in short time. This invention provides a system and methods for creating quick links for continuous content that are mostly desired by users. Context differentiated distributions of stop positions relative to the sections of interests when users are browsing continuous contents, for example, a web page or image, are calculated base on experiments. When users browse continuous content on production environment, the stop positions during their browsing, for example, the stop positions on a web page for product details page, are collected. The stop time, the device types of users, and the browsing context of users are also collected. The system then determines for the sections of interests of users based on the statistics of overall stop positions of all users of the same device and context, by matching the statistics against the previous studied distributions of stop positions for the same kind of device and context. And then the system could generate quick links automatically by summarizing content of interests determined by the system.

Advantage: Accurately locating sections of interests in continuous content. Generation of quick links for sections of interests in continuous content without human interference. After determining the sections of interests, these sections could be used to better improve content, or extracting keywords for the categories of the content. Step of the method:

1. Experiments are done to study the distributions of stop positions relative to the sections of content that are browsed when users stop scrolling. For example for a product details web page with a most interested section of product size, when the user is using a small screen for example mobile, the distribution could be pretty sharp, while the distribution generated by users using desktop could be relative smooth. (Figure 1 & Figure 2) A distribution density could be calculated to produce normalized probabilities. For example the distribution density for each location could be total stop time at that position of all users, normalized by the total stop time of all users for the current web page.

2

Figure 1: 3 sample stop positions when users browse the size section on a web page using a small screen. And a sample distribution of stop positions for all experimented users.

3

Figure 2: Sample distribution of stop positions for all experimented users who use a large screen.

2. On production environment, the distributions of user stop positions on each content for each kind of device and content are collected and calculated. The context when the data is produced are also collected. The context data include device type, resolution, or any of other behavior context. The distributions from...