Vamshi Ambati

Syntax-Driven Learning of Sub-Sentential Translation Equivalents and Translation Rules from Parsed Parallel Corpora

We describe a multi-step process for automatically learning reliable sub-sentential syntactic phrases that are translation equivalents of each other and syntactic translation rules between two languages. The input to the process is a corpus of parallel sentences, word-aligned and annotated with phrase-structure parse trees. We first apply a newly developed algorithm for aligning parse-tree nodes between the two parallel trees. Next, we extract all aligned sub-sentential syntactic constituents from the parallel sentences, and create a syntax-based phrase-table. Finally, we treat the node alignments as tree decomposition points and extract from the corpus all possible synchronous parallel tree fragments. These are then converted into synchronous context-free rules. We describe the approach and analyze its application to Chinese-English parallel data.

Collaborative workflow for crowdsourcing translation

In this paper we explore the challenges in crowdsourcing the task of translation over the web in which remotely located translators work on providing translations independent of each other. We then propose a collaborative workflow for crowdsourcing translation to address some of these challenges. In our pipeline model, the translators are working in phases where output from earlier phases can be enhanced in the subsequent phases. We also highlight some of the novel contributions of the pipeline model like assistive translation and translation synthesis that can leverage monolingual and bilingual speakers alike. We evaluate our approach by eliciting translations for both a minority-to-majority language pair and a minority-to-minority language pair. We observe that in both scenarios, our workflow produces better quality translations in a cost-effective manner, when compared to the traditional crowdsourcing workflow.

Digitizing a million books: challenges for document analysis

This paper describes the challenges for document image analysis community for building large digital libraries with diverse document categories. The challenges are identified from the experience of the on-going activities toward digitizing and archiving one million books. Smooth workflow has been established for archiving large quantity of books, with the help of efficient image processing algorithms. However, much more research is needed to address the challenges arising out of the diversity of the content in digital libraries.

A collaborative filtering based re-ranking strategy for search in digital libraries

Users of a digital book library system typically interact with the system to search for books by querying on the meta data describing the books or to search for information in the pages of a book by querying using one or more keywords. In either cases, a large volume of results are returned of which, the results relevant to the user are not often among the top few. Re-ranking of the search results according to the users interest based on his relevance feedback, has received wide attention in information retrieval. Also, recent work in collaborative filtering and information retrieval has shown that sharing of search experiences among users having similar interests, typically called a community, reduces the effort put in by any given user in retrieving the exact information of interest. In this paper, we propose a collaborative filtering based re-ranking strategy for the search processes in a digital library system. Our approach is to learn a user profile representing users interests using Machine Learning techniques and to re-rank the search results based on collaborative filtering techniques. In particular, we investigate the use of Support Vector Machines(SVMs) and k-Nearest Neighbour methods (kNN) for the task of classification. We also apply this approach to a large scale online Digital Library System and present the results of our evaluation.

An Improved Statistical Transfer System for French-English Machine Translation

This paper presents the Carnegie Mellon University statistical transfer MT system submitted to the 2009 WMT shared task in French-to-English translation. We describe a syntax-based approach that incorporates both syntactic and non-syntactic phrase pairs in addition to a syntactic grammar. After reporting development test results, we conduct a preliminary analysis of the coverage and effectiveness of the systems components.

Statistical Transfer Systems for French-English and German-English Machine Translation

We apply the Stat-XFER statistical transfer machine translation framework to the task of translating from French and German into English. We introduce statistical methods within our framework that allow for the principled extraction of syntax-based transfer rules from parallel corpora given word alignments and constituency parses. Performance is evaluated on test sets from the 2007 WMT shared task.

Improving re-ranking of search results using collaborative filtering

Search Engines today often return a large volume of results with possibly a few relevant results. The notion of relevance is subjective and depends on the user and the context of search. Re-ranking of these results to reflect the most relevant results to the user, using a user profile built from the relevance feedback has proved to provide good results. Our approach assumes implicit feedback gathered from a search engine query logs and learn a user profile. The user profile typically runs into sparsity problems due to the sheer volume of the WWW. Sparsity refers to the missing weights of certain words in the user profile. In this paper, we present an effective re-ranking strategy that compensates for the sparsity in a user’s profile, by applying collaborative filtering algorithms. Our evaluation results show an improvement in precision over approaches that use only a user’s profile.

Proactive Learning for Building Machine Translation Systems for Minority Languages

Building machine translation (MT) for many minority languages in the world is a serious challenge. For many minor languages there is little machine readable text, few knowledgeable linguists, and little money available for MT development. For these reasons, it becomes very important for an MT system to make best use of its resources, both labeled and unlabeled, in building a quality system. In this paper we argue that traditional active learning setup may not be the right fit for seeking annotations required for building a Syntax Based MT system for minority languages. We posit that a relatively new variant of active learning, Proactive Learning, is more suitable for this task.