We propose a novel Thai grapheme-tophoneme conversion method based on a neural regression model that is trained using neural networks to predict the similarity between a candidate and the correct pronunciation.After generating a set of candidates for an input word or phrase using the orthography rules, this model selects the best-similarity pronunciation from the candidates.This method can be applied to languages other than Thai simply by preparing enough orthography rules, and can reduce the mistakes that neural network models often make.We show that the accuracy of the proposed method is .931,which is comparable to that of encoder-decoder sequence models.We also demonstrate that the proposed method is superior in terms of the difference between correct and predicted pronunciations because incorrect, strange output sometimes occurs when using encoder-decoder sequence models but the error is within the expected range when using the proposed method.

Inputting query terms on a mobile terminal is frustrating because of device limitations though mobile web search is becoming popular. Query prediction is a promising approach for mobile search. In the literature, however, little attention has been paid to the first query term, though there are many reports on query expansion or second query recommendation. In this paper, we propose a first query prediction method that enables users to search related information for the current browsed webpage with an easier interface such as touch operation without any background search processes. The proposed method consists of body-text extraction, candidate query term extraction, and a scoring process. We also implemented a one-touch search application that works for Japanese webpages at a practical speed on a smartphone. According to our closed evaluation with 299 webpages and open evaluation with 14 users, our method achieved practical quality.

This poster presents an overview of the characteristics of a one-button information retrieval interface with closed captions from TV watching activities, which is intended to lighten the burden of remembering and entering query terms while watching TV. We investigated this interface with an experimental system named Video Bookmarking Search, which estimates query terms from closed captions with named-entity recognition and sentence labeling techniques. According to an empirical evaluation for 1,138 search queries from 206 bookmarks using seven actual TV shows on city life, travel, health, and cuisine, we found wider queries and search results are acceptable through the query-input-free interface, despite the fact that the number of queries and search results that are directly relevant to the users' original intentions is not high. The main reason is a watching user's interest is wider than what is expressed with query terms.