Integrating large-scale web data and curated corpus data in a search engine supporting German literacy education

摘要

Reading material that is of interest and at the right level for learners is an essential component of effective language educa-tion. The web has long been identified as a valuable source of reading material due to the abundance and variability of materials it offers and its broad range of attractive and current topics. Yet, the web as source of reading material can be problematic in low literacy contexts. We present ongoing work on a hybrid approach to text retrieval that combines the strengths of web search with retrieval from a high-quality, curated corpus re-source. Our system, KANSAS Suche 2.0, supports retrieval and rcranking based on criteria relevant for language learning in three different search modes: unrestricted web search, filtered web search, and cor-pus search. We demonstrate their comple-mentary strengths and weaknesses with re-gard to coverage, readability, and suitabil-ity of the retrieved material for adult lit-eracy and basic education. Wc show that their combination results in a very versa-tile and suitable text retrieval approach for education in the language arts.