• LarkcProject
• WP2
• WP2PluginWorkshop

Venue and attendance

The WP2 plugin workshop will be on Monday 23 March in Sheffield.

Please confirm your attendance at:

1. the workshop
2. restaurant meal booked for evening of 23 March

What you need to bring

1. A checked out LarKC trunk, compiled, and with the example running.
   • Please email WP5 if you have difficulty doing this
2. Your own code, in a runnable state, prefereably also checked in to the LarKC repository

Travel details

The workshop will start at 9:30 am in Room G30 on the ground floor of the Department of Computer Science at the University of Sheffield.

GATE travel page: http://www.gate.ac.uk/travel/

Directions to the Department: http://www.shef.ac.uk/visitors/

Official University visitor pages: http://www.shef.ac.uk/dcs/travel.html

Internet access

Wireless access will be available:

• eduroam for those from institutions that support it
• guest wireless service, key available on the day

Preliminary agenda

• 9:30 start
• Introduction
• 9:45 coffee
• Short technical presentations with code examples, approx 20 minutes each:
  • Coding to the LarKC plugin API (Mark Greenwood)
  • Using the LarKC data layer (Damyan Ognianoff)
  • Running plugins on the LarKC platform (Blaz Fortuna)
• Break out into pairs and groups along the lines of the M12 deliverables, for pair and team programming
  • 13:00 working lunch
  • Continue breakouts, rearranging as necessary
  • 15:15 coffee
  • Continue breakouts, rearranging as necessary
• 17:00 Report back
• 17:30 end

Suggested groups for breakouts

Here a a few suggestions: please add others, or we can discuss on the day.

1. Installing, running, and compiling against LarKC Although most attendees should already have the LarKC platform up and running, with their code checked in, it is possible that some will not. We may therefore have a breakout that deals with this.

2. Wrapping Semantic Annotation as a LarKC plugin We have nearly completed a plugin for this, and it could make a good example to show how to wrap existing code as a plugin. This group will look at completing the work.

3. Wrapping the MPG plugin If needed, a couple of people should look at this.

4. Running WP7b use case plugins on the platform USFD has writen a couple of plugins for the WP7b use case. These compile, how do we now run them on the platform?

5. WICI's application for granular selection in Semantic Literature Search WICI has developed a prototype system on "Computer Science Literature Search" based on our proposed idea for granular selection. We need suggestions on how our granular selection component could be plugged into the LarKC pipeline well, and we want to get some comments on how future version of the prototype system can grow into a good use case for "Selection and Retrieval" using LarKC. WICI's introduction on related programs can be found in here: WICI_Program_Introduction.pdf.

Meal

• Time: 19:30

• Place: Platillos

• Directions: on google maps

Confirmed attendance

Report

It was generally agreed that the workshop had been a useful opportunity to get plugin developers and platform developers together. Most participants had useful face-to-face communications with others. In the late morning and throughout the afternoon, we split the workshop into four informal groups, working on separate plugin-related topics. These are reported on below.

USFD: Running WP7b use case plugins on the platform

We had already written plugins for the WP7b cancer research use case, but had never run these in the platform. The brief for this subgroup was to implement a platform demonstrator, along the lines of the BabyLarKC or the Urban City demonstrator. This was successfully done, with a very simple single plugin instantiation of the platform.

• Issue - the current plugin could most usefully split into 3. The level of plugin granularity expected by the platform is smaller than that provided in this plugin. This will be done in the next phase.

USFD: Wrapping Semantic Annotation as a LarKC plugin

The intention was to wrap USFD's GATE toolkit, enabling an arbitrary GATE application to be run over documents, transforming them into semantic annotations encoded as triples. A plugin was completed, and a very basic single plugin platform written to test it. Several issues were encountered:

• Some details of the plugin are currently hard coded (e.g. the GATE application path). These will be passed as part of a Context object, to be provided in the next release of the API

• The document to be analysed is hard coded. This will be passed as a ?NaturalLanguageDocument, to be provided in the next release of the API

• There were some difficulties with library conflicts between GATE and LarKC. These were overcome by loading all libraries via the start up shell script, LarKC first and then GATE, so that LarKC versions get priority in the classloader.
• We were unable to print Statements for debugging, the print giving a NPE in either OpenRDF code or ORDI code.

WICI: Receiving suggestions on granular selection application and a use case

WICI has developed one application as a use case of granular selection: user centered computer science literature search system, and detailed report can be seen in here WICI_Program_Introduction.pdf. During the workshop, Zhisheng, Luka, Jose, Blaz have provided many valuable suggestions for us.

• Zhisheng mentioned that we haven't consider the semantic distance of keywords which are used for queries, for example, "reasoning" and "logic" are treated as two isolated keywords in user interests and queries, and we will work on that in the near future.
• Luka suggested that user interest centered semantic query refinement can be a good example that related to Quality of Service(QoS) in WP5, hence, we are going to have a discussion with WP5 later about this and seek for possible cooperation.
• Luka and Blaz suggested that query refinement based on user interest can be passed to reasoner plug-in to handle, or many other approaches can be tried if possible.
• Jose suggested a comparison of WICI's work on the RDF file for user interest with Application Profile Markup Language(APML) and we may have future cooperation consider triple set sub-setting in RDF graph, since MPG and WICI are all working on network statistics for selection from different perspectives.

We also get many new information concerning the change of current API from Damyan and Luka, and we would like to thank Mark and Angus for two valuable plug-in examples worked out from this workshop.

VUA: Reasoner plugin

With the helps of Damyan (?OntoText) and Luka (?CycEuro), Zhisheng (VUA) fixed some problems of the Pellet SPARQLDL reasoner plug-in with the LarKC platform v0.4. After the workshop, the Pellet SPARQLDL Reasoner has been released on the LarKC platform v0.4.

Comments from WP5 members attending the workshop

(to be added)