Browse Prior Art Database

Automatic quiz generator from social information gathered from eBooks, videos, and online training Disclosure Number: IPCOM000246147D
Publication Date: 2016-May-12
Document File: 5 page(s) / 78K

Publishing Venue

The Prior Art Database


Training materials are spread over digital media including articles and videos delivered through the internet. Online training makes it very convenient for enterprises to broadcast their training content, often with interactive quizzes to get the feedback from readers. Content and the questions are rarely personalized, or if they are, the personalization amounts to picking a survey or quiz from among a small finite set that is authored in advance. No matter what culture, region, educational background and basic knowledge a user comes to the training with, users often get the same questions in their training materials.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 30% of the total text.

Page 01 of 5

Automatic quiz generator from social information gathered from eBooks, videos, and online training

In an educational setting, it is difficult to generate tests which will assess a student's knowledge across both student selected readings or learning materials and instructor selected materials.

In an enterprise setting, a similar situation occurs in confirming an employee's self reported educational tasks.

Much of time, the content of the training is only suitable for some employees. Here are some cases:

Some users already learned related knowledge before, and the quizzes are too easy for them. User don't want to browse all content and answer pointless


quizzes, it is wasting their time. They hope to get deeper questions and get something new in this training.

The quiz may be too difficult for other employees, they have not gathered enough basic knowledge before the training.


Sometimes users are compelled to click on links in training materials in order to receive a certification or spend a certain amount of time on a page whose


material they have already mastered, wasting the user's time (and an employer's money).

For the training publisher or organization, the feedback from the quiz doesn't differentiate those who have really learned something. For the training receiver, they are unable to demonstrate their knowledge and may not focus on the next logical learning priorities for their level of skill.

A well known technique in quiz generation is to vary the difficulty of questions later in the test based on a user's responses early in the test. For instance, if the user gets the first 5 questions right, the quiz becomes harder. If they get the first 5 questions wrong, the quiz becomes easier. In these cases, overall and comparative scoring usually takes into account the difficulty of the questions given to the user. This method assumes the user's level of competency is consistent across a corpus and answers to one question are predictive of the user's strength across all questions.

With this invention, the digital training content can be broadcast anywhere, but the content is both convertible and precise. The does not just deliver or select training or quizzes from a predefined set, but generates quizzes for each individual using analysis of social data and learning material consumption.

The techniques for each step are related as discussed below, and each step contains novel techniques, particularly:

Novels ways to use crowd-sourced data from multiple readers to find areas to focus on

Using crowd-sourced data to map questions to different profile groups based on responses within the profile group and content consumption behaviors from the group
Novel ways to identify specific questions for an individual based on the individual's consumption of the content and quizzes
Automatic generation or recommendations of expertise profiles based on areas skipped or skipped where the user's mastery is confirmed through the automated qui...