On the use of finite state transducers for semantic interpretation

Raymond, Christian; Béchet, Frédéric; Mori, Renato De; Damnati, Géraldine

doi:10.1016/j.specom.2005.06.012

Cited by 46 publications

(43 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…As its name implies, it uses multiple language models for ASR, such as finite-state-grammars and N-gram statistical models, and multiple models for language understanding, such as finite-state-transducers [7], weighted finite-state-transducers [8], [9], keyphrase extractors, and conditional-random-field-based models.…”

Section: Conceptmentioning

confidence: 99%

Automatic Allocation of Training Data for Speech Understanding Based on Multiple Model Combinations

Komatani

Nakano

Katsumaru

et al. 2012

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYThe optimal way to build speech understanding modules depends on the amount of training data available. When only a small amount of training data is available, effective allocation of the data is crucial to preventing overfitting of statistical methods. We have developed a method for allocating a limited amount of training data in accordance with the amount available. Our method exploits rule-based methods for when the amount of data is small, which are included in our speech understanding framework based on multiple model combinations, i.e., multiple automatic speech recognition (ASR) modules and multiple language understanding (LU) modules, and then allocates training data preferentially to the modules that dominate the overall performance of speech understanding. Experimental evaluation showed that our allocation method consistently outperforms baseline methods that use a single ASR module and a single LU module while the amount of training data increases. key words: spoken dialogue system, language understanding, rapid prototyping, limited amount of training data

show abstract

Section: Conceptmentioning

confidence: 99%

Automatic Allocation of Training Data for Speech Understanding Based on Multiple Model Combinations

Komatani

Nakano

Katsumaru

et al. 2012

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…Thus we approach SLU by reranking the hypotheses generated by a baseline model: in our case we use two different local models, i.e. SFSTs [7] and CRF [2]. Our discriminative reranking is modeled with SVMs, which also enable the use of kernelbased learning [8].…”

Section: Introductionmentioning

confidence: 99%

Discriminative Reranking for Spoken Language Understanding

Dinarelli

Moschitti

Riccardi

2012

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Abstract-Spoken Language Understanding (SLU) is concerned with the extraction of meaning structures from spoken utterances. Recent computational approaches to SLU, e.g. Conditional Random Fields (CRF), optimize local models by encoding several features, mainly based on simple n-grams. In contrast, recent works have shown that the accuracy of CRF can be significantly improved by modeling long-distance dependency features. In this paper, we propose novel approaches to encode all possible dependencies between features and most importantly among parts of the meaning structure, e.g. concepts and their combination. We rerank hypotheses generated by local models, e.g. Stochastic Finite State Transducers (SFSTs) or Conditional Random Fields (CRF), with a global model. The latter encodes a very large number of dependencies (in the form of trees or sequences) by applying kernel methods to the space of all meaning (sub) structures. We performed comparative experiments between SFST, CRF, Support Vector Machines (SVMs) and our proposed discriminative reranking models (DRMs) on representative conversational speech corpora in three different languages: the ATIS (English), the MEDIA (French) and the LUNA (Italian) corpora. These corpora have been collected within three different domain applications of increasing complexity: informational, transactional and problemsolving tasks, respectively. The results show that our DRMs consistently outperform the state-of-the-art models based on CRF.

show abstract

“…There are a lot of works using weighted word graphs as intermediate representation of uttered sentences in automatic speech recognition (ASR) systems or in spoken language understanding (SLU) systems [1][2][3][4][5][6][7]. Some authors use the concept of word lattice and other ones use the concept of word confusion network (WCN).…”

Section: Introductionmentioning

confidence: 99%

“…A SLU system where the output of the ASR module is a lattice of word hypotheses is described in [5]. Finite state transducers (FST) are used in a translation process for generating hypotheses of conceptual constituents from the lattice of word hypotheses.…”

Section: Introductionmentioning

confidence: 99%

Using Word Graphs as Intermediate Representation of Uttered Sentences

Gómez

Sanchís

2012

Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications

View full text Add to dashboard Cite

Nodes represent discrete instants of time, arcs are labelled with words, and a confidence measure is assigned to each detected word, which is computed by using the phonetic probabilities of the subsequence of acoustic frames used for completing the word. We evaluated the obtained word graphs by searching the path that best matches with the correct sentence and then measuring the word accuracy, i.e. the oracle word accuracy.

show abstract

On the use of finite state transducers for semantic interpretation

Cited by 46 publications

References 22 publications

Automatic Allocation of Training Data for Speech Understanding Based on Multiple Model Combinations

Automatic Allocation of Training Data for Speech Understanding Based on Multiple Model Combinations

Discriminative Reranking for Spoken Language Understanding

Using Word Graphs as Intermediate Representation of Uttered Sentences

Contact Info

Product

Resources

About