Weekly Technical Meeting - WP3 - 19th October 2011

Today Sahithi and Eamonn met for one hour to discuss the progress of Work Package three of eRaUI. We discussed a number of areas related to the interface of the widget and related functionality. Firstly, we agreed that the scenario screens which are to be developed, giving details of use cases of eRaUI and associated screenshots, is to be postponed until some further development work is completed. We also agreed that in relation to testing and evaluating eRaUI, it would require at least another two to four weeks of development before we are at a stage to effectively do this.

We looked at the use of live-stream user activity monitoring which has been implemented on the eRaUI admin user interface. We discusse the possibility of displaying miniaturised user sessions within iFrames, which would be an exciting and novel way to present user activity. However we agreed that there are a number of technical limitations which would make it difficult to achieve, viz. the lack of an effective cross-browser technique to miniaturise a webpage within an iFrame (currently only IE supports this to any extent) and also the performance issues which would entail from displaying so many user sessions simultaneously.

We discussed the current weighting system used to delineate terms into categories (i.e. Novice, Student, Expert) and agreed that it is better to make this an automatically generated system - using machine-learning algorithms to automatically assign weightings to each category. We will of course have a means of manually overriding these settings - and this has been built in to the current admin interface of eRaUI.

We discussed the admin-user live communication facility and decided it would be excellent to have peer-peer chat ability enabled also. We discussed some novel ways of presenting user activity to the user. This would create the impression of a virtual community on the site.

WP3 – Meeting with Sahithi Siva on 12th October 2011

On the 12^th October 2011 Eamonn and Sahithi met to discuss the progress on Work Package 3. We discussed the means by which we can generate different scenarios for the use of eRaUI, viz.

• · Highlighting and greying of page elements in relation to relevance to user level
• · Search results displayed within the widget.
• · Administrator intervention to guide the user remotely
• · A combination of the above facilities

We agreed that it is important to define the visual aspects of the user interface of eRaUI. Primarily, we agree that the widget will occupy the right-hand space of any page upon which it is deployed while the existing page content (i.e. a copy of the nactem site in its current form) will reside on the left-hand side.

We also plan this week to generate scenario screens for different uses of eRaUI.

WP2 Meeting 13th October: Designing & implementing Machine learning algorithms, workflow based interface and content-based algorithms

On the 13^th October 2011 Farhi Marir and Eamonn met to discuss progress regarding Work Package 2 deliverables. We began with a demonstration of the present functionality of eRaUI, including the new ontology editing tools which have been created, whereby administrators can edit terms within a dictionary (ontology) which are matched to weightings indicating their specificity to a category (i.e. novice, student, expert). We also saw a demonstration of the new ‘term highlighting and hiding’ proposed by Sahithi, whereby users are presented with the content most suited to their requirements in highlighted form, while less relevant content is greyed out.

Within the new editable ontology terms are incremented automatically according to frequency of clicking. This is a provisional system, and it has been proposed that this is made admin-changeable only. We discussed a simple provisional algorithm for determining the overall category of a term, whereby the component weightings are divided by the sum of all components to give an ‘average relevance’. Furthermore, we discussed the use of ‘dynamic categories’, whereby administrators are able to decide which types of user can be present in the system (i.e. novice, expert, student or other). This makes eRaUI into a more generic tool than it would be otherwise. We decided that for dynamically created categories, the administrator must be able to specify a sort order for these categories to appear in.

The widget is designed to attempt to determine the user’s category (i.e. level of expertise) while they are browsing. While the widget displays the predicted level and allows this to be changed, we agree that this shall be kept for informational purposes only in later prototypes.

We also saw a rudimentary demonstration of the real-time user stream, whereby the administrator can detect and examine user sessions in real-time. We agreed that we should be able to demarcate users currently using the system by user level and country, to facilitate analysis and administrator intervention if necessary.

Lastly we discussed the use of window panes, whereby we can display items of content related to search terms entered into the widget. We noted that we must create a new ontology of terms to support this mechanism. For instance, searching for ‘Seminars’ might also yield ‘workshops’ and ‘conferences’, which will be placed into their own window panes as a preview.

Technical Meeting - WP4 – Notes for blog – Meeting with Yanguo Jing on 10th October 2011

In terms of test and evaluation, we discussed the user of PhD students, researchers in or outside of Londonmet. JISC may help to recommend people in the NaCTeM community who may want to participate in the evaluation. We also discuss the measurements need to be established and to be used to measure the effectiveness of eRaUI. A comparative evaluation study will be carried out to evaluate the usefulness of eRaUI. We discussed also the means by which eRaUI has been tailored for easy integration with any website by means of adding a single line of code. Thus we have ensured that integration is considered right from the outset of the project.

Technical Meeting - WP3 – Meeting with Sahithi Siva on 5th October 2011

On the 5th October 2011 Sahithi and Eamonn met to discuss progress so far, and consider some new ideas which we might implement within eRaUI. First of all, we saw a demonstration of the new DOM element-relative user behaviour tracking, which has been designed to greatly improve the accuracy of tracking versus the previous system which was page-relative. We viewed the action of eRaUI user behaviour tracking on the current NaCTeM feedback form, and saw how it is possible to track user keypresses.

We discussed producing some new video demonstrations of the current functionality of eRaUI including a tour of the mechansism by which data is recorded into the eRaUI central database. These shall be available on the eRaUI blog shortly.

We considered a couple of different ways to broaden the functionality of eRaUI in such a way as to make the interface upon over which it is deployed more adaptive to user needs. The first method involves manipulating the DOM of the web page currently being viewed in such a way as to change the visual precedence of items which are likely to be of more interest to the user, based upon their current interaction with the interface so far. For instance, elements which are unlikely to be of interest could be obscured completely while elements of likely interest could be made larger or highlighted. This can all be achieved dynamically via javascript manipulation of DOM elements.

The second method considered to make the interface more adaptive is to allow an element of administrator-user interaction via the widget. Administrators could be made able to view users' interactions with the interface in real-time and assist those users who might be in need of help, for instance by means of a text chat area. This is standard practice in many commercial websites already, however eRaUI tracking capability could make it far easier for administrators to become aware of the precise nature of users' difficulty in navigating or using an interface for a specific task. Mouse pointer emulation in real-time could be used to allow adminstrators to demostrate to users how to perform specific tasks.

We also discussed the use of window panes with eRaUI in order to present information which the user might be looking for in a cascading window format. The precise form this will take is yet to be established.

We have therefore decided upon four separate ways in which it would be possible to enhance the adaptability of a user interface so as to serve the needs of individual users. These are:

- suppression and highlighting of elements with the DOM of the interface web-page according to user-relevance.

- facilitation of administrator-user assistance by means of text conversation and mouse emulation

- suggestion of relevant links and autocomplete capability of the eRaUI widget search box

- A combination of all of the above approaches.

It has been established, therefore, that current development of eRaUI will focus upon the prototype implementation of the above 4 mechanisms of enhancing interface adaptability. Once completed to prototype level, we shall attempt to discover the viability and usefulness of each of these options by getting users to try them.

Technical Meeting - 6th October - WP2: Designing & implementing Machine learning algorithms, workflow based interface and content-based algorithms

We discussed the implementation of two main WP2 functionalities of eRAUI: machine learning algorithms to model the behaviour of user and content-based algorithm to support the user in his search.

First of all, we saw Eamonn demonstration of the new DOM element-relative user behaviour tracking, which has been designed to greatly improve the accuracy of tracking versus the previous system which was page-relative. We viewed the action of eRaUI user behaviour tracking on the current NaCTeM feedback form, and saw how it is possible to track user keypresses. We discussed producing some new video demonstrations of the current functionality of eRaUI including a tour of the mechanism by which data is recorded into the eRaUI central database. These shall be available on the eRaUI blog shortly. Farhi asked Eamonn to consider the implementation of ontology of terms which would be used to match commonly searched-for keywords and the anchor text of links to levels of user expertise. Since the anchor text of links clicked on within NaCTeM are recorded automatically in the back-end database, these can be stored in a dictionary which an administrator can then classify according to expertise level. This will facilitate the classification of users, for instance, into novice, expert and PhD level. During the meeting, Eamonn was able to prototype and demonstrate the workability of this feature. Farhi suggested that the user's level of expertise is not explicitly alterable from within the interface of the eRaUI widget - instead a system of internal classification - viz. that of the ontology mentioned above - shall be used instead.

We also discussed the use of window panes with eRaUI in order to present information which the user might be looking for in a cascading window format. The precise form this will take is yet to be established.

Eamonn suggested collaboration between E-research admin and individual users. We also decided that we shall give a presentation of the current functionality of eRaUI to the faculty of computing at London Metropolitan University on the 2nd November 2011. Further to this, we have got some of our MSc students involved in the project with the CPP103N Software Engineering module (Large Scale OODB Development). These students are currently working upon aspects of eRaUI from a software engineering context.

Technical Meeting - 6th October WP1: Project management & Dissemination

On the 6th October 2011 Farhi (leader of WP1 & WP2) & Eamonn met to discuss progress of WP1 progress.

• Farhi informed Eamonn that there will separate weekly WP meeting involving at least the leader of WP and the research associate. Eamonn will be reporting the minutes of each WP meeting on the Blog.

• Farhi & Eamonn will prepare the first presentation (out of three planned during the next two months) to FoC seminar on the 2nd of November to get feedback from the faculty researchers on two WP2 functionalities: implementation of some machine learning algorithms to model user behaviour and content and collaborative techniques to support the user in his/her search.

• Farhi reported that some parts of eRAUI work packages are given as a coursework to his MSc Software Engineering students taking the module CPP103N (Large Scale OODB Software Development). These students are currently working upon aspects of eRaUI from within a software engineering context.

• Farhi & Eamonn also discussed about the organisation of the first steering committee meeting and it was decided the meeting will be happening at the end of November or the beginning of December 2011. In this meeting eRaUI team will be reporting on the progress of each WP and demonstrating the first eRaUI prototype to get feedback from the steering committee.

We discussed the proposed implementation of an ontology of terms which would be used to match commonly searched for keywords and the anchor text of links to levels of user expertise. Since the anchor text of links clicked on within NaCTeM are recorded automatically in the back-end database, these can be stored in a dictionary which an administrator can then clasify according to expertise level. This will facilitate the classification of users, for instance, into novice, expert and PhD level.

We also discussed making it so that the users level of expertise is not explicitly alterable from within the interface of the eRaUI widget - instead a system of internal classification - viz. that of of the ontology mentioned above shall be used instead.

eRaUI weekly technical meeting - Thursday 29th September

The eRaUI team met on Thursday 29th September to discuss the progress so far. We saw a demonstration of the means by which key presses can now be captured and played back using the eRaUI interaction playback tool. We also discussed the details of the revisions which have been made to the mechanisms of capture for user interactions with the web interface. Previously all user interaction captures were made relative to the document body, however this causes issues with different browsers and platforms which render elements in different positions. Much of the work this week has involved revising the mechanism of capture to work in a way which is relative to individual DOM elements. This means, among other things, altering the DOM of each document so as to give a unique ID attribute to every element which is to be rendered.