The 3rd Workshop on Making Sense of Microposts (#MSM2013) was held in Rio de Janeiro, Brazil, on the 13th of May 2013, as part of the 22nd International Conference on the World Wide Web (WWW'13). #MSM2013 is the third in a series of successful workshops. The #MSM workshop was rst held at the 8th Extended Semantic Web Conference (ESWC 2011), and with approximately 50 participants, was the most popular workshop at ESWC 2011. The second workshop was held at the 21st International Conference on the World Wide Web (WWW'12), and had approximately 60 participants, as did this year's workshop.
The #MSM series of workshops is unique in targeting both Semantic Web researchers and other elds, within Computer Science, such as Human-Computer Interaction and Visualisation, and in other areas, particularly the Social Sciences. The aim is to harness the benets dierent elds bring to research involving Microposts. The workshop also encourages the demonstration of the generation and use of Microposts through dierent physical and online media, as well as application of research, and re-use of Micropost data in real-world scenarios. Continuing to hold the workshop at WWW allows us to reach a wider and more varied audience and target research and applications at the leading edge of technology. The 2013 edition was an occasion to expand our community, and with the conference in Rio de Janeiro, to connect with local researchers from Brazil and South America, opening the way for new synergy and interesting discussions within the local cultural context.
In a world where more and more data is becoming available to machines, questions related to the use of this data for increasing machine intelligence naturally arise. Big Data treatment eorts exploit masses of data using statistical approaches in order to conceive anticipatory systems able to predict future human behaviour and adapt to it. Semantic analysis of Web content, including Microposts, is another complementary perspective to the goal of making machines more intelligent and more capable of supporting daily human activity, decision making and communication. We are seeing a very large increase in systems relying on Semantic Web technologies being deployed: Intelligent Assistants, such as Siri1 , rely on Semantic Data Graphs to provide users with factual responses to their questions. Facebook Graph Search2 allows users to formulate complex queries over a socio-semantic graph constructed from people's likes and structural knowledge about things being liked. While static knowledge bases are largely employed in such systems, exploiting the dynamic, evolving knowledge that resides in the growing masses of Microposts, invaluable as they are acknowledged to be, remains a major challenge.
Each year we make a little step toward resolving this challenge, due largely to what makes publishing via Microposts so popular their brevity, and as a result, the use of non-standard abbreviations, informal language and grammar.
With each workshop we have found that our research community continues to open exciting new possibilities for constructing increasingly intelligent and useful services.
New to this year's workshop is the Concept Extraction Challenge, sponsored by eBay. Existing concept extraction tools are intended for use over news corpora and similar document-based corpora with relatively long length. The aim of the challenge was to foster research into novel, more accurate concept extraction for (much shorter) Micropost data. The keen interest in concept extraction that is shared by our community motivated this challenge, focused for this rst time on a rather general task. The interest shown in the challenge by both academia and industry has conrmed its relevance. We aim to pursue the challenge in the future editions of #MSM, and are investigating new challenge tasks and the use of dierent collections of data, prompted by the challenge results and further research it continues to foster.
This rst run of the challenge has been a learning curve, with contributions from participants, not just in their formal submissions, but also to corrections in the training data that fed into the cycle of updates that resulted in the nal gold standard. The #MSM2013 Concept Extraction Challenge received 22 complete submissions, out of which 6 were accepted for presentation at the workshop, and a further 7 for presentation as posters. Submissions came from institutions across 12 countries, with 13% of submitting authors from Brazilian institutions.
Many hearty thanks to all our contributors and participants, and also the Programme Committees whose valued feedback resulted in a rich collection of work, each of which adds to the state of the art in leading edge research in the challenging task of information extraction from Microposts. Especial thanks to Andrea Varga, who was largely responsible for generating the challenge dataset, and Pablo Mendes who gave us very useful suggestions on collaborative annotation of the data. We are condent that the #MSM series of workshops will continue to foster a vibrant community, and target the rich body of information generated by the many and varied authors whose social and working lives span the physical and online worlds.
This volume includes rst a challenge report, with a summary of the state of the art and a comparison of the performance of the approaches taken for the 13 submissions accepted. This provides an overview of the capability of the state of the art in Concept Extraction approaches to date. This introductory paper details the challenge objectives and task, and the dataset construction and validation processes. We also provide a comprehensive description of the quantitative evaluation methodology followed and the performance and ranking metrics used.
Participants' descriptions of the systems implemented complete the proceedings. Each submission was peer reviewed, to provide the authors with feedback on their approach and to identify interesting and promising work to present at the workshop. The quantitative evaluation described in the report was also carried out to rank submission runs this was the nal criterion, with a cut-o for acceptance, and the key measure for the challenge award.
The call for participation and all challenge abstracts, in addition to those for the main workshop track, are available on the #MSM2013 website 7 . The full challenge proceedings are also available on the CEUR-WS server, as Vol-1019 8 . The proceedings for the main track are available as part of the WWW'13 Proceedings Companion 9 . The proceedings for the 1st and 2nd workshops are available as CEUR Vol-718 10 and Vol-838 11 respectively.