Cross-language Learning-to-Rank for Patent Retrieval, Phase 2: Weakly Supervised Learning of Cross-lingual Systems

• Applicant: Professor Dr. Stefan Riezler
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing ; General and Comparative Linguistics, Experimental Linguistics, Typology, Non-European Languages
• Term: from 2012 to 2019
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 211613886

• Final Report Year: 2018

Final Report Abstract

Effective information search across languages is a key problem in today’s information society. For example, cross-lingual patent prior art search is an important tool to determine a patent’s novelty and to avoid patent infringement. For high accuracy, machine learning approaches require costly manual annotation of supervision signals such as relevance links across languages for crosslingual retrieval. We could show that cross-lingual rankings can be learned directly from data that are weakly supervised, but are not strictly parallel. Such weak supervision signals can be relevance indicators such as citations in patents or hyperlinks in Wikipedia pages. Our project showed that similar techniques can be successfully applied to optimize cross-lingual retrieval and to train machine translation systems on massive non-parallel data.

Publications

• (2012). Joint feature selection in distributed stochastic learning for large-scale discriminative training in SMT. In Proc. of the Meeting of the Association for Computational Linguistics (ACL), Jeju Island, South Korea
  Simianer, P., Riezler, S., and Dyer, C.
  
• (2012). Structural and topical dimensions in multi-task patent translation. In Proc. of the Conference of the European Association for Computational Linguistics (EACL), Avignon, France
  Wäschle, K. and Riezler, S.
  
• (2013). Boosting cross-language retrieval by learning bilingual phrase associations from relevance rankings. In Proc. of the Conference on Empirical Methods in Natural Language Processing (EMNLP), Seattle, Washington, USA
  Sokolov, A., Jehl, L., Hieber, F., and Riezler, S.
  
• (2013). Generative and discriminative methods for online adaptation in SMT. In Proc. of the Machine Translation Summit (MTSummit), Nice, France
  Wäschle, K., Simianer, P., Bertoldi, N., Riezler, S., and Federico, M.
  
• (2014). Learning translational and knowledge-based similarities from relevance rankings for cross-language retrieval. In Proc. of the Association of Computational Linguistics (ACL), Baltimore, USA
  Schamoni, H., Hieber, F., Sokolov, A., and Riezler, S.
  (See online at https://dx.doi.org/10.3115/v1/P14-2080)
• (2014). Online adaptation to post-edits for phrase-based statistical machine translation. Machine Translation, 28:309–339
  Bertoldi, N., Simianer, P., Cettolo, M., Wäschle, K., Federico, M., and Riezler, S.
  (See online at https://doi.org/10.1007/s10590-014-9159-7)