Galke L, Gerstenkorn G & Scherp A (2018) A Case Study of Closed-Domain Response Suggestion with Limited Training Data. In: Elloumi M, Granitzer M, Hameurlain A, Seifert C, Stein B, Tjoa A & Wagner R (eds.) Database and Expert Systems Applications. DEXA 2018. Communications in Computer and Information Science, 903. DEXA 2018: International Conference on Database and Expert Systems Applications, 03.09.2018-06.09.2018. Cham, Switzerland: Springer International Publishing, pp. 218-229. https://doi.org/10.1007/978-3-319-99133-7_18
Abstract We analyze the problem of response suggestion in a closed domain along a real-world scenario of a digital library. We present a text-processing pipeline to generate question-answer pairs from chat transcripts. On this limited amount of training data, we compare retrieval-based, conditioned-generation, and dedicated representation learning approaches for response suggestion. Our results show that retrieval-based methods that strive to find similar, known contexts are preferable over parametric approaches from the conditioned-generation family, when the training data is limited. We, however, identify a specific representation learning approach that is competitive to the retrieval-based approaches despite the training data limitation.