The increasing availability of text information coded in many di erent languages poses new challenges to modern information retrieval and mining systems in order to discover and exchange knowledge at a larger world-wide scale. The 1st International Workshop on Modeling, Learning and Mining for Cross/Multilinguality (dubbed MultiLingMine 2016) provides a venue to discuss research advances in cross-/multilingual related topics, focusing on new multidisciplinary research questions that have not been deeply investigated so far (e.g., in CLEF and related events relevant to CLIR). This includes theoretical and experimental on-going works about novel representation models, learning algorithms, and knowledge-based methodologies for emerging trends and applications, such as, e.g., cross-view cross-/multilingual information retrieval and document mining, (knowledge-based) translation-independent cross/multilingual corpora, applications in social network contexts, and more.
In the last few years the phenomenon of multilingual information overload has received signi cant attention due to the huge availability of information coded in many di erent languages. We have in fact witnessed a growing popularity of tools that are designed for collaboratively editing through contributors across the world, which has led to an increased demand for methods capable of effectively and e ciently searching, retrieving, managing and mining di erent language-written document collections. The multilingual information overload phenomenon introduces new challenges to modern information retrieval systems. By better searching, indexing, and organizing such rich and heterogeneous information, we can discover and exchange knowledge at a larger world-wide scale. However, since research on multilingual information is relatively young, important issues still remain uncovered: { how to de ne a translation-independent representation of the documents across many languages; { whether existing solutions for comparable corpora can be enhanced to generalize to multiple languages without depending on bilingual dictionaries or incurring bias in merging language-speci c results; { how to pro tably exploit knowledge bases to enable translation-independent preserving and unveiling of content semantics; { how to de ne proper indexing and multidimensional data structures to better capture the multi-topic and/or multi-aspect nature of multi-lingual documents; { how to detect duplicate or redundant information among di erent languages or, conversely, novelty in the produced information; { how to enrich and update multi-lingual knowledge bases from documents; { how to exploit multi-lingual knowledge bases for question answering; { how to e ciently extend topic modeling to deal with multi/cross-lingual documents in many languages; { how to evaluate and visualize retrieval and mining results. 2
The aim of the 1st International Workshop on Modeling, Learning and Mining for Cross/Multilinguality (dubbed MultiLingMine 2016 ),6 held in conjunction with the 2016 ECIR Conference, is to establish a venue to discuss research advances in cross-/multilingual related topics. MultiLingMine 2016 has been structured as a full-day workshop. Its program schedule includes invited talks as well as a panel discussion among the participants. It is mainly geared towards students, researchers and practitioners actively working on topics related to information retrieval, classi cation, clustering, indexing and modeling of multilingual corpora collections. A major objective of this workshop is to focus on research questions that have not been deeply investigated so far. Special interest is devoted to contributions that aim to consider the following aspects: { Modeling: methods to develop suitable representations for multilingual corpora, possibly embedding information from di erent views/aspects, such as, e.g., tensor models and decompositions, word-to-vector models, statistical topic models, representational learning, etc. { Learning: any unsupervised, supervised, and semi-supervised approach in cross/multilingual contexts. { The use of knowledge bases to support the modeling, learning, or both stages of multilingual corpora analysis. { Emerging trends and applications, such as, e.g., cross-view cross-/multilingual
IR, multilingual text mining in social networks, etc.
Main research topics of interest in MultiLingMine 2016 include the following: { Multilingual/cross-lingual information access, web search, and ranking 6 http://events.dimes.unical.it/multilingmine/ { Multilingual/cross-lingual relevance feedback { Multilingual/cross-lingual text summarization { Multilingual/cross-lingual question answering { Multilingual/cross-lingual information extraction { Multilingual/cross-lingual document indexing { Multilingual/cross-lingual topic modeling { Multi-view/Multimodal representation models for multilingual corpora and crosslingual applications { Cross-view multi/cross-lingual information retrieval and document mining { Multilingual/cross-lingual classi cation and clustering { Knowledge-based approaches to model and mine multilingual corpora { Social network analysis and mining for multilinguality/cross-linguality { Plagiarism detection for multilinguality/cross-linguality { Sentiment analysis for multilinguality/cross-linguality { Deep learning for multilinguality/cross-linguality { Novel validity criteria for cross-/multilingual retrieval and learning tasks { Novel paradigms for visualization of patterns mined in multilingual corpora { Emerging applications for multilingual/cross-lingual domains
The ultimate goal of the MultiLingMine workshop is to increase the visibility of the above research themes, and also to bridge closely related research elds such as information access, searching and ranking, information extraction, feature engineering, text mining and machine learning. 3
The scienti c signi cance of the workshop is assured by a Program Committee which includes 20 research scholars, coming from di erent countries and widely recognized as experts in cross/multi-lingual information retrieval: Ahmet Aker, Univ. She eld, United Kingdom Rafael Banchs, I2R Singapore Martin Braschler, Zurich Univ. of Applied Sciences, Switzerland Philipp Cimiano, Bielefeld University, Germany Paul Clough, Univ. She eld, United Kingdom Andrea Esuli, ISTI-CNR, Italy Wei Gao, QCRI, Qatar Cyril Goutte, National Research Council, Canada Parth Gupta, Universitat Politcnica de Valncia, Spain Dunja Mladenic, Jozef Stefan International Postgraduate school, Slovenia Alejandro Moreo, ISTI-CNR, Italy Alessandro Moschitti, Univ. Trento, Italy; QCRI, Qatar Matteo Negri, FBK - Fondazione Bruno Kessler, Italy Simone Paolo Ponzetto, Univ. Mannheim, Germany Achim Rettinger, Institute AIFB, Germany Philipp Sorg, Institute AIFB, Germany Ralf Steinberger, JRC in Ispra, Italy Marco Turchi, FBK - Fondazione Bruno Kessler, Italy
Vasudeva Varma, IIIT Hyderabad, India Ivan Vulic, KU Leuven, Belgium 4
A COLING'08 workshop [
The main event related to our workshop is the CLEF initiative [
Our intention is to continue the lead of previous events about multilingual related topics, however from a broader perspective which is relevant to various information retrieval and document mining elds. We aim at soliciting contributions from scholars and practitioners in information retrieval that are interested in Multi/Cross-lingual document management, search, mining, and evaluation tasks. Moreover, di erently from previous workshops, we would emphasize some speci c trends, such as cross-view cross/multilingual IR, as well as the growing tightly interaction between knowledge-based and statistical/algorithmic approaches in order to deal with multilingual information overload.