Abstract

With the increasing democratization of elec-tronic media, vast information resources areavailable in less-frequently-taught languagessuch as Swahili or Somali.That informa-tion, which may be crucially important andnot available elsewhere, can be difficult formonolingual English speakers to effectivelyaccess. In this paper we present SARAL, anend-to-end cross-lingual information retrieval(CLIR) and summarization system for low-resource languages that 1) enables Englishspeakers to search foreign language reposito-ries of text and audio using English queries,2) summarizes the retrieved documents in En-glish with respect to a particular informa-tion need, and 3) provides complete transcrip-tions and translations as needed. The SARALsystem achieved the top end-to-end perfor-mance in the most recent IARPA MATERIALCLIR+summarization evaluations.

Details