adenu research group implementing the support for adaptation in openacs/.lrn current works at adenu...

31
aDeNu Research Group http://adenu.ia.uned.es Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

Upload: warren-lynch

Post on 28-Dec-2015

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group http://adenu.ia.uned.es

Implementing the support for adaptation in OpenACS/.LRN

Current works at aDeNu group

Page 2: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 2/31

Contents

• A combined presentation of the papers:– Dynamic support in OpenACS/.LRN: Technological

infrastructure for providing dynamic recommendations for all in open and standard-based LMS

• Olga Santos, Emmanuelle Raffenne, Jorge Granado, Jesús G. Boticario (UNED, Spain)

– A General Tracking and Auditing Architecture for the OpenACS framework (See abstract)

• Jorge Couchet, Olga C. Santos, Emmanuelle Raffenne, Jesús G. Boticario, Daniel Manrique (ORT/UNED/UPM, Uruguay/Spain)

– Management of standard-based User Model and Device Profile in OpenACS (See abstract)

• Adrián Cuartero, Olga C. Santos, Jorge Granado, Emmanuelle Raffenne y Jesús G. Boticario (UNED, Spain)

Page 3: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 3/31

Dynamic support in OpenACS/.LRN

• Objective• Background• Why OpenACS• Approach• The prototype• Ongoing works

Page 4: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 4/31

Dynamic Support in OpenACS/.LRN

• Objective:– provide dynamic support to the user in an inclusive

way at the course execution to overcome impasses that learners may encounter, and which are not covered by the design of the course. To provide this support, the learner’s interactions and their evolution overtime are considered.

• Recommending service– Under development at aDeNu group

• Following the IMS standards (LIP, accLIP, QTI, LD, ...)• Based on user modelling techniques (rules-based or/and

machine learning based)– Technical infrastructure integrated into

OpenACS/.LRN via web services.

Page 5: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 5/31

Background

• EU4ALL project– there is a need for modelling user requirements and

providing adapted responses to them • to “accommodate the diversity of ways people

interact with technology and the content and services it delivers” (DoW Obj)

• to “offer their teaching and services in a way that is accessible to all learners including those that are sometimes described as disabled learners” (DoW obj.)

• aLFanet & ADAPTAPlan projects– Universal design approaches is not enough. There is

a need for dynamic support at runtime that considers the learners’ interactions and their evolution over time.

Page 6: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 6/31

Why OpenACS

• We require:– An LMS that supports the delivery of contents,

activities and services following the appropriate instructional design and fostering collaboration

– Accessibility and adaptation support• Although the current state of the art in learning

environments shows that there is no environment that fulfil these requirements, our experience shows that OpenACS/.LRN is the most suitable platform

Page 7: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 7/31

Our approach: Theory

• Application of user modelling techniques to provide adaptation– storing user information (accLIP/ISO) and applying

the appropriate adaptation strategies considering users' preferences and the context in the LMS

• Recommender technology to provide dynamic support to learners during the course execution in an inclusive way – The information from other people is used to select,

filter, sort or highlight items to the user

Page 8: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 8/31

Our approach: Practice

• The Accessible and Adaptive Module (A2M) – uses artificial intelligence techniques – pervasive use of standards (accLIP, LD, MD, WAI, WS):

• to model learners and resources accessibility features• to follow the course design• communicate with other systems

• The objective of the A2M is twofold: 1. Update the user models from the learners’ interactions using

machine learning techniques.2. Generate dynamic contextual recommendations during the

course execution based on these models and collaborative filtering (similar actions -Amazon-) and collaborative content techniques (similar content)

Page 9: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 9/31

Our Approach for Modelling

• Apply and relate existing specifications (LIP etc.) to promote reusability among systems (e.g. LMS -Moodle-)

• Three types of user characteristics are considered to generate the adaptation:

– Felder learning styles: defines several dimensions regarding how people process information

– The knowledge competency level (per course objective) based on Bloom’s taxonomy, whose improvement is measured through IMS-QTI questionnaires

– The collaborative competency level (per course): computed with AI techniques (rules o machine learning), which take into account the usage of the course services (forums, file-storage, etc.).

• Device capabilities and the accessibility preferences of the users are also considered:

– IMS LIP (Learner Information Package) and AccLIP to model the learner and the preferences for alternative resources

– CC/PP (with UAProf vocabulary) to model the device capabilities• CC/PP: Composite Capability/Preference Profiles (howto define a device)

• UAProf: User Agent Profile (vocabulary that can be used in CC/PP to define mobile devices caracteristics)

Page 10: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 10/31

Approach for the recommendations

• Hybrid approach– Part based on knowledge:

• users like you liked ... (based on ratings)• people who did this also did …• users like you (based on personal attributes)

– Part based on learning:• Based on the Utility of an object for the user• Rules based inferences: rules from existing

relations between items and users' needs

Page 11: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 11/31

The prototype

• Offers a recommendation service to be used by external components to ask for the appropriate recommendations for the user currently working in the LMS.

• This first prototype is focused on the integration with the LMS via web services communication and the persistence of the data

• To access the required information, parallel developments are used:– An OpenACS/.LRN package to compute the learning style of the

user (Felder's index of learning style). – An external component which offers web services to manage the

learner and the device profiles (details later): accLIP, LIP, CC/PP– Tracking and Auditing Engines infrastructure (TAE package) that

is being built in OpenACS (details later)

Page 12: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 12/31

Client side

The recommendation service (OpenACS package):• Relies on xo-soap (SOAP protocol plugin for xorb).

Thanks to Stefan Sobernig :-)

• A portlet displays the result from the request. • The response consists on:

– an introductory text– a list of recommendations

• Process– Calls sent when page loaded– Response received and displayed– User is free to follow the recommendation– If recommendation followed, message is sent to the service to

inform that the action has been taken

Page 13: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 13/31

The recommendations

• Each recommendation is a piece of text (message) describing a possible action to be done by the user.

• Part of the text may be a link• The link is defined by:

– the content: the text that is shown to the user and consists on the underlying part of the recommendation offered

– the title: the information saying where the link goes (i.e. the title attribute of the A tag)

– the pointer: the URI that opens the link, it can be a URL or an object identifier within the LMS, depending on the type

– the type: it can be internal to the LMS (an object) or an external URL

Page 14: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 14/31

Snapshot

Page 15: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 15/31

Server Side

• Implements the basic infrastructure of the A2M (Accessible and Adaptive Module)

– Currently: integration with the LMS and persistence of the data• Technology:

– JDK1.5 to be run on Apache Tomcat– Apache Axis2 is used as web service engine.

• Two processes:– Online

• the front-end: waits for a request from the client• the back-end (under development): produces the appropriate

recommendations for each user taking into account her user model, the context, past interactions and psychopedagogical guidelines

– Offline• the attributes of the models are learnt

Page 16: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 16/31

On going works

• Focused on the back-end and the off line process of the A2M (Accessible and Adaptative Module)

– intensive use of AI techniques required to offer dynamic support in LMS

• These tasks are being undertaken within the scope of Olga C. Santos Ph.D thesis, supervised by Jesús G. Boticario (UNED)

Page 17: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 17/31

A General Tracking and Auditing Architecture for the OpenACS

framework

• The packages• Tracking and Auditing processes• The architecture• Applications• Open Issues

Page 18: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 18/31

TAE Packages

• TAE: Tracking and Auditing Engines• Provides a general tracking and auditing

architecture• OpenACS only offers ad-hoc solutions:

– ClickStream (WEG?)– User Tracking (Telefonica for e-LANE)

Page 19: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 19/31

Tracking and Auditing processes

Page 20: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 20/31

TAE architecture for OpenACS

Page 21: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 21/31

Architecture components 1/2

• Tracking Engine: It hides to the Auditing Engine (and the TAE user) the recording mechanism. The engine receives the definition of the data (raw data and metadata) that needs to be recorded, and creates automatically the filters needed.

• Auditing Engine: It offers a centralized access and administration to the TAE, and defines a Service Contract for the Auditing Engine (SCae) in order to distribute the audit function among the different OpenACS packages, each one with its own information.

Page 22: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 22/31

Architecture components 2/2

• SCae Implementation for the audited packages:Each OpenACS package implements the SC– defining the objects to be tracked and audited– providing the semantic processing functions in order to

extract the meaning from the data recorded by the Tracking Engine.

• Report Engine (Jasper): It hides to the Auditing Engine (and the TAE user) the complexities of the report creation. The engine receives:– the data source definition (XML) – the report configuration (XML)

Page 23: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 23/31

Applications

• Previous background:– Jorge and Daniel research is to develop new machine learning

algorithms in order to be used to extract useful knowledge from big datasets. (e.g.: from access logs)

• Mining .LRN user’s needs:– By combining auto-organized systems and learning design

specifications, we intend to mine, analyze and categorize the sequence of actions followed by learners to provide dynamic recommendations that improve the efficiency and efficacy of their learning.

Page 24: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 24/31

Open issues

• Best mechanism for the recorded data growth?– there could be two solutions:

• limit what will be tracked on each record?• setting a periodic process that archives the log data to a

separate file after extracting the relevant information?

• The recording process itself:– storing directly into the database?– or having an intermediate step?

Page 25: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 25/31

Management of standard-based User Model and Device Profile in

OpenACS

• Objectives• User Model• Device Profile• Application in the EU4ALL Project

Page 26: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 26/31

Objectives

• Solution to provide an adapted and inclusive support in LMS– Supports:

• IMS-LIP, IMS-AccLIP – the combination of these two specifications define the learner

information profile, considering accessibility preferences

• CC/PP & UAProf– CC/PP model the device capabilities and UAProf provide a

vocabulary for it

• Based on an initial proposal presented in Vienna ('07 spring conference): Modelling Learner Interaction Preferences

Page 27: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 27/31

User Model server (LIP, accLIP)

• getUserModelAttribute• setUserModelAttribute• deleteUserModelAttribute• getUserModel• setUserModel• deleteUserModel

Page 28: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 28/31

Device Profile server (CC/PP + UAProf)

• getDeviceProfile• setDeviceProfile• deleteDeviceProfile• getDeviceProfileParameter• setDeviceProfileParameter

Page 29: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 29/31

Application in EU4ALL Project

• User Model + Device Profile --> Content Personalization

• User Model:– profile created/sent when the user register at the platform– user can edit her personal information

• preferences• learning style (Felder)

• Device Profile:– captured by a proxy during a request– or filled by the user

• Content Personalization:– requests info to User Model and Device Profile– adapts the response according to this information

Page 30: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 30/31

Conclusion

• Work to provide dynamic support in .LRN to improve the learner experience

• Pervasive use of IMS specifications (LIP, accLIP, CC/PP, etc.)

• Apply AI techniques (rules based and machine learning)

• Rely on gathering interactions data in the LMS

Page 31: ADeNu Research Group  Implementing the support for adaptation in OpenACS/.LRN Current works at aDeNu group

aDeNu Research Group 31/31

Thank You