Báo cáo khoa học: "Learning to Rank" doc

Ngày tải lên: 23/03/2014, 17:20

... different sources to take into account the difference in the importance of terms depending on the text. We also used sentences, sections (for Wikipedia arti- cles only) and documents as units to calculate doc- ument ... annotators annotated rankings for each of the 150 people individually. The annota- tors were instructed to rank the definitions assuming that they were creating a “who...

Ngày tải lên: 08/03/2014, 03:20

... similarity-based models are doomed to perform poorly for QA be- cause they fail to “bridge the lexical chasm” be- tween questions and answers. One way to address this problem is to learn question -to- answer trans- formations ... designed originally for factoid answer validation and does not adapt well to our problem. Because the length of non-factoid answers is typically significant...

Ngày tải lên: 31/03/2014, 00:20

... between pareto vs. non-pareto points ignores the fact that 2nd-place non-pareto points may also lead to good practical solutions. A better approach may be to adopt a graded definition of Pareto optimality ... weight vector w, takes in a foreign sen- tence f and returns a translated hypothesis h. The argmax operates in vector space and our goal is to find w leading to hypotheses on the P...

Ngày tải lên: 19/02/2014, 19:20

... Thus, we also associate features to each Chinese token (characters or words) to reflect the likelihood of the token being part of the translation. We describe how to train a CRF model for identifying ... potentially be used to assist translators to find the most common translation for a given term, or to supplement a bilingual terminology bank (e.g., adding multilingual titl...

Ngày tải lên: 19/02/2014, 19:20

... that the key to maximum dialogue ef- fectiveness is to listen to the user. This paper de- scribes the development of an adaptive dialogue sys- tem that uses the feedback of users to automatically improve ... harder to detect than others, the system’s classification accuracy depends largely on the user’s barrier. To prevent classification accuracy distorting the user evaluations, we...

Ngày tải lên: 20/02/2014, 12:20

... MT requires taking additional factors into consideration, compared to the monolingual task described above. On the one hand, we need to extend our model to handle the addi- tional knowledge ... than in the Kyoto Corpus, because our bunsetsu-parsing algorithm prefers to err on the side of making too many rather than too few phrases. This is because our final goal is to generate...

Ngày tải lên: 23/03/2014, 18:20

... a data-driven approach to story generation that does not require extensive manual involvement. Our goal is to create stories automatically by leverag- ing knowledge inherent in corpora. Stories within the ... wish to extract and quantify. Of course, build- ing a database of characters and their actions is merely the first step towards creating an automatic story generator. The latter mus...

Ngày tải lên: 30/03/2014, 23:20

... it is her task to describe the path to the instruction follower, who cannot see the reference path. Our system learns to interpret these navigational direc- tions, without access to explicit linguistic ... θ T φ(s, a) most closely approximates E[R(s, a)]. To learn these weights θ we use SARSA (Sutton and Barto, 1998), an online learn- ing algorithm similar to Q-learning (Watkins a...

Ngày tải lên: 20/02/2014, 04:20

... for ease of reference to any line in this doc- ument. We will use this document to illustrate the details of our learning approach throughout this pa- per. We refer to a horizontal line as ... 12: 13: Net tons Capability 14: produced utilization 15: Week to March 14 1,633,000 75.8~ 16: Week to March 7 1,570,000 71.9~ 17: Year to date 15,029,000 66.9~ 18: Year earlier t...

Ngày tải lên: 20/02/2014, 19:20