Project Overview

A large-scale On-line Social Network supporting efficient adaptive learning methodologies. The project utilizes crowdsourcing techniques, collaborative filtering and collective intelligence, dynamic data warehousing, data analysis and data mining techniques.

That allows gaining educational knowledge via adaptive learning through teaching and adaptively converging to the most productive learning pathways with respect to a particular group of students and their performance profiles. Users of the system contribute concise and clear lessons focused on a very specific topic. Each lesson has four mandatory components:

  • a lesslet name, referring to the concept it is going to teach
  • a clear and concise explanation of the concept
  • an example illustrating the explanation in a simple and intuitive way
  • a test, including a small number of questions, to assess users' success on the lesslet
The test part of a lesslet allows us to keep track of the specific user's performance and to build a weighted directed graph, where vertices are the lesslets and edges show direction of users' leaps on them. The weight of the edge shows estimated success of transferring from one lesslet to another and is calculated based on a popularity of the lesslet and the average success ratio of all users who took it. The weight allows the system to make a recommendation for a user for following steps.

Here you can see the presentation of the SALT and what users can do in it (the presentation has slide animations, however it doesn't have automatic transitions between slides, so if you see that nothing is going on on the slide, you can go to next one).

Similar and Related Works

–Goal: Allow user to learn and teach how to code in interactive way.
–Dis: focused only on coding, no personalization.
–SALT: interactive programming environment is planned for interactive examples and tests
–Goal: world-wide study group. More like Q&A site.
–Dis: no personalization
–SALT: user can create wishes (a kind of item Q&A), however wishes are transformed into lesslets. Personalization for wishes to respond.
–Goal: Suite of products designed as Learning management system. Could be used to organized classes
–Dis: No crowdsourcing, no personalization
•Peer 2 Peer University
–Goal: Teaching and learning by peers for peers. Allow anyone to create courses.
–Dis: Users are expected to create/take whole course, not personalization, no online collaboration.
–Goal: Teach and learn courses. Course recommendation.
–Dis: Users are expected to create/take whole course. Some courses are not free, not much crowdsourcing.
–Goal: Create recommendation engine for education.
–Dis: no crowdsourcing and social networking. Seems like courses are to be created by professors and system is intended to be used in class.
–Goal: To provide a free world-class education for anyone anywhere. Video lectures. Interactive examples. Provide suggestions.
–Dis: no crowdsourcing, no social networking.
•Other MOOC systems (e.g., Coursera) provide video lectures recorded by professors. Users are supposed to take whole classes. No personalization.

Research Topics

The SALT project touches on a number of important data research topics, including application of recommender systems in online education, complex adaptive systems, educational data mining, user reliability and trust, and more.

System Implementation

SALT is implemented as a web application, therefore it requires no installation and accessible anywhere via internet and user favorite web browser. At the time of writing this paragraph, SALT was not open to the general public yet. We schedule the release on soon. However, you still can try to access it here. Please see some screenshots below as well as the tools, systems, libraries, technologies we use. Basically all the implementation is also shown in action in the presentation above.
System Screen Shots (you can click on any image to enlarge)

Related Publications

  • Self-Adaptive Learning through Teaching.
    Evgeny Karataev, Vladimir Zadorozhny
    iFest 2012
    Remarks: Not peer reviewed.

    Fast development of computers, mobile devices, internet, web and social networks changed the way people share information on-line. Each user may produce and consume a substantial amount of data. However, the ease and convenience of sharing data poses new problems; systems for sharing data become decentralized and loosely coordinated, which makes the search for a piece of information by a particular user difficult.

    This project addresses the problem of adaptive information processing in a large scale decentralized and loosely coordinated systems based on crowdsourcing and social computing. Such systems can be thought of as complex adaptive systems (CAS). Two very interesting properties of the CAS, emergence and self-organization, allow global structure or patterns to appear in a system without a central authority or external element imposing it through planning. However, there is question whether such systems can produce any useful information, self-organize and adapt for users. Perhaps emergence and self-organization will not converge to produce desired properties.

    As an example of CAS, we propose to build a complex adaptive information system as an on-line social network to improve the process of learning. Therefore, in addition to the research contribution, our project will bring a solid practical result to engage students in an active learning process. It will help more students (especially those who are underprepared) excel in various courses by utilizing their growing familiarity with social media.

In preparation or under review
  • Adaptive Social Learning Based On Crowdsourcing.
    Evgeny Karataev, Vladimir Zadorozhny
    ICDCS 2014
    Remarks: Full paper

    We introduce a novel social learning paradigm that adapts to needs of individual users based on their collective learning experiences. The proposed approach combines concepts of Crowdsourcing, Online Social Networks and Complex Adaptive Systems to engage users in efficient learning through teaching. The process of adaptive convergence is facilitated through dynamic data analysis techniques that we implemented in the SALT (Self-Adaptive Learning through Teaching) system. Our results show that collective learning experiences can be efficiently utilized in adaptive social learning.

Team Members and Affiliations

Vladimir Zadorozhny
(Associate Professor @ SIS , Pitt )
Evgeny Karataev
(Doctoral Student @ SIS , Pitt )

Communication, Q&A

Please feel free to send me email if you have any comments or questions.


Project page last time updated on Jan. 21, 2014

© Evgeny Karataev 2018

Last Modified on November 2nd, 2016