Báo cáo khoa học: "Learning to Generate Naturalistic Utterances Using Reviews in Spoken Dialogue Systems" ppt

... Association for Computational Linguistics Learning to Generate Naturalistic U tterances Using Reviews in Spok en Dialogue Systems Ryuichiro Higashinaka NTT Corporation rh@cslab.kecl.ntt.co.jp Rashmi ... on which Minipar was trained, the output of Minipar can be inaccurate, leading to failure in conversion. We check whether conversion is suc- cessful in the filtering stage....

Ngày tải lên: 23/03/2014, 18:20

... operat- ing in a stochastic environment, and optimise it using Reinforcement Learning. The in- put to the module is a Communicative Goal supplied by the Dialogue Manager. The CG consists of a Dialogue ... evaluate a new model for NLG in Spoken Dialogue Systems as planning under uncertainty. In Section 2 we argue for applying RL to NLG problems and explain the overall fra...

Ngày tải lên: 08/03/2014, 21:20

... scoring first in the Information Gain rank, validating our central hypothesis. The testing of the classifier using each of the above feature sets was performed on the remain- ing 25% of the Town-Info ... by the In- formation Gain measures that were calculated on the training set using WEKA, as shown in table 3. 8.1 Information Gain Quite surprisingly, we note that the rank given by th...

Ngày tải lên: 31/03/2014, 20:20

... easily be plugged in to test their usefulness in different dialogue contexts or for new domains. 4 Training the Dialogue System 4.1 Random strategy generation In its training mode, the dialogue system ... similar to ours, constisting of: generation of (limited) exploratory dialogue data, using a train- ing system; creating an MDP from these data and the rewards assigned by...

Ngày tải lên: 20/02/2014, 12:20

... combines two distinct clauses into a single sentence with a coordinating or subordinating conjunction (e.g. Above has decent decor BUT Carmine’s has good decor), while cue-word insertion inserts a cue ... domains. For example, the training method used in SPoT (Sentence Planner Train- able), as described in (Walker, Rambow and Ro- gati, 2002), was only shown to work in the travel domai...

Ngày tải lên: 23/03/2014, 19:20

... learning. 4.3 Training Ranking Models Using the reference ranking data, we trained a rank- ing model using the ranking SVM (Joachims, 2002) (with a linear kernel) that minimizes the pairwise ranking ... 0.13. Finally, taking the means of ranks given to each definition, we merged the indi- vidual rankings to create the reference rankings. An example of a reference ranking is the bottom...

Ngày tải lên: 08/03/2014, 03:20

... information is added to the training set T (line 8), which is then optimized by any conventional subroutine in line 10. We will follow PRO in using a pairwise classifier in line 10, which finds w ∗ that ... necessarily pareto-optimal. Algorithm 1 shows the pseudocode. In line 3, we take a point h ∗ and check if it is dominating or dominated in the for- loop (lines 4-8). At least o...

Ngày tải lên: 19/02/2014, 19:20

... while requiring minimal human intervention to prepare the training data. 3 Method To find translations for a given term on the Web, a promising approach is automatically learning to extract ... 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 titles to ex...

Ngày tải lên: 19/02/2014, 19:20

... set- ting it is situated in, motivating automated learn- ing approaches to acquiring meaning. Tradi- tional accounts of learning typically rely on lin- guistic annotation (Zettlemoyer and Collins, ... our attention to just the utterances of the instruction giver, ignoring the instruction follower. This is to reduce redundancy and noise in the data - the in- struction follower rar...

Ngày tải lên: 20/02/2014, 04:20

... this paper due to space constraint. 443 Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line Line 1234567890123456789012345678901234567890123456789012345678901234567890 ... tables in Figure 1 to 3. Instead of resorting to an ad-hoc method to rec- ognize tables, we present a new approach in this pa- per that l...

Ngày tải lên: 20/02/2014, 19:20