Search

This paper presents a lightweight machine learning model and a fast conjunction matching method to the problem of identifying user intents behind their spoken text commands. These model and method were integrated into a mobile virtual assistant for Vietnamese (VAV) to understand what mobile users mean to carry out on their smartphones via their commands. User intent, in the scope of our work, is an action associated with a particular mobile application. Given an input spoken command, its application will be identified by an accurate classifier while the action will be determined by a flexible conjunction matching algorithm. Our classifier and conjunction matcher are very compact in order that we can store and execute them right on mobile devices. To evaluate the classifier and the matcher, we annotated a medium-sized data set, conducting various experiments with different settings, and achieving impressive accuracy for both the application and action identification.

Near-duplicate documents exacerbate the problem of information overload. Research in detecting near-duplicates has attracted a lot of attention from both industry and academia. In this paper, we focus on addressing this problem for Vietnamese documents which, to the best of our knowledge, has not been done before. Most of the current algorithms have been designed for English which are not directly applicable to Vietnamese - a monosyllabic language. We propose to combine Charikarpsilas algorithm (Alexander Andoni et al., 2006) with a ldquoweighting schemerdquo and Vietnamese specific features to address the language intricacy. Experimental results indicate that our scheme is effective for detecting near-duplicates in a corpus of Vietnamese documents.