SRILM Toolkit & Dialog Systems


Notes:

SRILM (SRI Language Modeling Toolkit) is an open source, extensible language modeling toolkit. SRILM is a C++-based toolkit for language modeling. Language models are built and interpolated using the SRILM. SRILM can be used for building local language models. SRILM is used to estimate n-grams Language Models (LM). SRILM has an API for computing word language model probabilities. Disambig is one module in SRILM. Perplexity values can be computed with SRILM. There is a standard script for “compute – best – mix” in the SRILM package. The LM weighted using SRILM has been used to train language models. With an SRILM extension, efficient estimation of maximum entropy language models with n-gram features can be achieved. Even with relatively small language models, SRILM can be used to prune the language models using an entropy criterion. N-gram models may be estimated for all of the possible combinations using SRILM. SRILM can be used to build n-gram ARPA format language models. SRILM reads and writes to a standard ARPA (Advanced Research Projects Agency) file format for n-gram models. Standard n-gram language models may be trained with the SRILM using interpolated modified Kneser-Ney smoothing. SRILM can be used to build bigram language models from various corpora, such as the English Gigaword corpus. SRILM can be used on a monolingual training corpus of 48,000,000 sentences, for example.

A bigram language model used in recognition systems was generated using the SRILM with the modified Kneser-Ney back-off discounting. Trigram LMs may be estimated using the SRILM employing the default Good-Turing discounting method. The language model is a capitalization-invariant tri-gram language model with Good-Turing discounting acquired from the training corpus using the SRI language modeling toolkit. Modified KN models may be estimated on training set count files and applied to the test set using SRILM. A 4-gram target LM with unmodified Kneser-Ney backoff discounting was generated using the SRILM. SRILM was used to train a 5-gram language model on the English sentences of FBIS (Foreign Broadcast Information Service) corpus. A 5-gram language model generated by the SRILM can be used in the cube-pruning process. VMM (variable memory modeling) may be implemented within SRILM and compared to default N-Gram models. SRILM can also be used to train a 7-gram model on training set. For instance, SRILM may be used to estimate individual language models for truthful and deceptive opinions. Translation models and generation models may be trained by the Moses toolkit. IRSTLM is another, similar language modeling toolkit. N-gram language models may be scored using z-scores. For example z-scores have been used to compare documents by examining how many standard deviations each n-gram differs from its mean occurrence in a large collection, or text corpus, of documents (which form the “background” vector).

Resources:

Wikipedia:

References:

See also:

IRSTLM (IRST Language Modeling) Toolkit | Kaldi ASR


tucSage: Grammar Rule Induction for Spoken Dialogue Systems via Probabilistic Candidate Selection A Chorianopoulou, G Athanasopoulou, E Iosif… – SemEval …, 2014 – anthology.aclweb.org … 1 Introduction A critical task for Spoken Dialogue Systems (SDS) is the understanding of the transcribed user input, that utilizes an underlying domain … Regarding the computation of perplexity-based features (defined in Section 2.1) the SRILM toolkit (A. Stolcke, 2002) was used. … Related articles All 9 versions

Using lexical, syntactic and semantic features for non-terminal grammar rule induction in spoken dialogue systems G Athanasopoulou, I Klasinas… – … (SLT), 2014 IEEE, 2014 – ieeexplore.ieee.org … The proposed system was evaluated with respect to two spo- ken dialogue system application domains: air travel and fi- nance (currency exchange … The computation of frequen- cies, probabilities and perplexities for each domain was com- puted using the SRILM toolkit [29]. … Cited by 1 Related articles

A Novel Similarity Measure to Induce Semantic Classes and Its Application for Language Model Adaptation in a Dialogue System YL Li, WQ Xu, YH Yan – Journal of Computer Science and Technology, 2012 – Springer Page 1. Li YL, Xu WQ, Yan YH. A novel similarity measure to induce semantic classes and its application for language model adaptation in a dialogue system. … Application for Language Model Adaptation in a Dialogue System … Related articles All 11 versions

Combining multiple translation systems for spoken language understanding portability F Garcia, LF Hurtado, E Segarra… – … (SLT), 2012 IEEE, 2012 – ieeexplore.ieee.org … Given the aforementioned characteristics of the limited- domain spoken dialog systems, it is possible to obtain manu- ally transcribed, segmented, and labeled corpora that can be used to learn statistical models for the … We used the SRILM toolkit for the estimations of the LMs. … Cited by 2 Related articles All 6 versions

Language model reduction for practical implementation in LVCSR systems S Ostrogonac, B Popovi?, M Se?ujski… – Infoteh, Jahorina, …, 2013 – infoteh.etf.unssa.rs.ba … This work was supported by the Ministry of Education, Science and Technological Development of Serbia within the Project “Development of Dialogue Systems in Serbian … [5] A. Stolcke, “SRILM – an extensible language modeling toolkit,” Proceedings of ICSLP, vol. … Cited by 7 Related articles All 4 versions

Using syntactic and confusion network structure for out-of-vocabulary word detection A Marin, T Kwiatkowski, M Ostendorf… – … (SLT), 2012 IEEE, 2012 – ieeexplore.ieee.org … ii, pp. II.21–II.24. [9] H. Sun et al., “Using word confidence measure for oov words detection in a spontaneous spoken dialog system,” in Proc. Eurospeech, 2003, pp. … 133 –138. [23] A. Stolcke, “SRILM – an extensible language modeling toolkit,” in Proc. ICSLP, 2002, pp. … Cited by 11 Related articles All 8 versions

Semantic parsing using word confusion networks with conditional random fields. G Tür, A Deoras, D Hakkani-Tür – INTERSPEECH, 2013 – msr-waypoint.net … Figure 3: Conceptual process of typical spoken dialog systems with cascaded speech recognition and understanding. … The word confusion networks are built using the SRILM toolkit [33], which uses a method similar to AT&T pivot algorithm [27]. … Cited by 12 Related articles All 8 versions

“Can you give me another word for hyperbaric?”: Improving speech translation using targeted clarification questions NF Ayan, A Mandal, M Frandsen… – … , Speech and Signal …, 2013 – ieeexplore.ieee.org … The complexity, ambiguity, and informality of natural spoken language have a significant negative impact on the overall per- formance of existing dialog systems. … For the fi- nal translation system, we generated a 6-gram language model us- ing the SRILM toolkit [25], on the target … Cited by 11 Related articles All 10 versions

Comparison and combination of lightly supervised approaches for language portability of a spoken language understanding system B Jabaian, L Besacier, F Lefevre – Audio, Speech, and …, 2013 – ieeexplore.ieee.org … through the availability of several toolkits (MOSES, Joshua, Jane etc.). … In dialogue systems, the SLU module has an intermediate role between ASR and DM. Its role is to … (which can be trained using the SRILM toolkit [38] on the corpus of concepts), and a translation model … Cited by 14 Related articles All 6 versions

Cache neural network language models based on long-distance dependencies for a spoken dialog system F Zamora-Martínez, S Espana-Boquera… – … , Speech and Signal …, 2012 – ieeexplore.ieee.org … Experiments conducted on a SLU task for a spoken dialog system have shown a significant CER re- duction using a rather simple model. … Interspeech, 2005. [14] A. Stolcke, “SRILM: an extensible language modeling toolkit,” in Proc. ICSLP, 2002, pp. 901–904. … Cited by 7 Related articles All 4 versions

ASR error detection using recurrent neural network language model and complementary ASR YC Tam, Y Lei, J Zheng, W Wang – Acoustics, Speech and …, 2014 – ieeexplore.ieee.org … request rephrasing or spelling of words) [1]. In this way, accurate ASR error detection can play a crucial role within a dialogue system. … We aligned the reference transcriptions against the confusion networks by using the SRILM toolkit [9] to obtain target labels for training and … Cited by 11 Related articles All 6 versions

Investigating Automatic & Human Filled Pause Insertion for Speech Synthesis R Dall, M Tomalin, M Wester… – Proceedings of the …, 2014 – homepages.inf.ed.ac.uk … 16] heard pairs of sentences with and without FPs and were asked whether the FP increased the naturalness of a voice for a dialogue system. … Random: Randomly inserts a single UH into a sentence 2. Ngram LM: A standard 4gram LM was built using the SRILM toolkit [36] and … Cited by 6 Related articles All 7 versions

Using web text to improve keyword spotting in speech A Gandhe, L Qin, F Metze, A Rudnicky… – … (ASRU), 2013 IEEE …, 2013 – ieeexplore.ieee.org … The interpolation weight was tuned on the development data using the “best-mix” script in the SRILM toolkit. … 1, pp. 573-576, 2005. [10] K. Weilhammer, MN Stuttle and S. Young, “Boot- strapping language models for dialogue systems,” Proc. ICSLP-2006, 2006. … Cited by 3 Related articles All 7 versions

Using multiple versions of speech input in phone recognition M Liberman, J Yuan, A Stolcke… – Acoustics, Speech and …, 2013 – ieeexplore.ieee.org … 10, pp. 89–99, 2002. [7] Bohus, D., Zweig, G., Nguyen, P., Li, X., “ Joint N-best rescoring for repeated utterances in spoken dialog systems,” Proceedings of SLT 2008, pp. … [11] Stolcke, A., SRILM – An Extensible Language Modeling Toolkit, Proceedings of ICSLP 2002, pp. … Cited by 1 Related articles All 15 versions

Prosodic and temporal features for language modeling for dialog NG Ward, A Vega, T Baumann – Speech Communication, 2012 – Elsevier … The specific model we used as the baseline is a standard order 3 (trigram) model, namely the one implemented in the SRILM toolkit (Stolcke, 2002), run with the default parameter settings, which include Katz’ back-off with Good–Turing discounting. … Cited by 15 Related articles All 15 versions

Language identification with limited resources E Sanchis, M Giménez, LF Hurtado – V Jornadas TIMM, 2014 – ceur-ws.org … This is the case of multilingual dialog systems where the system has to detect the input language in order to choose the … We used SRILM Toolkit [Sto02] to estimated the phonetic language models of the classifiers and HTK Speech Recognition Toolkit [You06] to perform the … Cited by 1 Related articles All 3 versions

Unsupervised language model adaptation for automatic speech recognition of broadcast news using web 2.0. T Schlippe, L Gren, NT Vu, T Schultz – INTERSPEECH, 2013 – csl.anthropomatik.kit.edu … 2-Pass Decoding Stategy With the help of the SRI Language Modeling Toolkit [25], we train individual 3-gram LMs … and T. Kawahara, “A Bootstrapping Approach for Devel- oping Language Model of New Spoken Dialogue Systems by Se … [25] A. Stolcke, “SRILM – An Extensible … Cited by 7 Related articles All 7 versions

Identification of code-switched sentences and words using language modeling approaches LC Yu, WC He, WN Chien, YH Tseng – Mathematical Problems in …, 2013 – hindawi.com … When dealing with code-switched input, intelligent systems such as dialog systems must be capable of identifying the various languages and recognize the speaker’s intention embedded in the input [7, 8]. However, it is a significant challenge for intelligent systems to deal with … Cited by 3 Related articles All 4 versions

Using Paraphrases and Lexical Semantics to Improve the Accuracy and the Robustness of Supervised Models in Situated Dialogue Systems C Gardent, LMR Barahona – Conference on Empirical Methods in Natural …, 2013 – hal.inria.fr … we used the S-Space Package http://code.google.com/p/ airhead-research/wiki/RandomIndexing 9. We used SRILM (http://www.speech.sri.com/ projects/srilm) … 2004. Data-driven strategies for an automated dialogue system. … Mallet : A ma- chine learning for language toolkit. … Related articles All 10 versions

Confidence Measures in Automatic Speech Recognition Systems for Error Detection in Restricted Domains J Olcoz, A Ortega, A Miguel, E Lleida – Advances in Speech and Language …, 2014 – Springer … its posterior probability P(phi) associated, obtained using the (lattice-tool) of the SRILM Toolkit [11 … original one, which has been created using the Finite State Machines (FSM) Toolkit [13 … R., Saz, O., Guijarrubia, V., Miguel, A., Torres, M., Lleida, E.: Improving dialogue systems in a … Related articles All 3 versions

Impacts of machine translation and speech synthesis on speech-to-speech translation K Hashimoto, J Yamagishi, W Byrne, S King… – Speech …, 2012 – Elsevier … The rest of this paper is organized as follows. Section 2 reviews related work on integrating natural language generation and speech synthesis for a single-language spoken dialog system and integrating machine translation and speech synthesis for S2ST. … Cited by 7 Related articles All 6 versions

Adapting Spoken Dialog Systems Towards Domains and Users M Sun – 2015 – cs.cmu.edu Page 1. February 13, 2015 DRAFT Adapting Spoken Dialog Systems Towards Domains and Users Ming Sun … Abstract Spoken dialog systems have been widely used across many domains. For exam- ple, voice applications are popular these days in smart devices. … Related articles

Deep Neural Network Based Continuous Speech Recognition for Serbian Using the Kaldi Toolkit V Delic – Speech and Computer: 17th International Conference, …, 2015 – books.google.com … corpus (6300000 tokens, consists of newspaper articles, books etc.), using the SRILM toolkit [15] and … DNN Based CSR for Serbian Using the Kaldi Toolkit 191 10.00 20.00 30.00 … the Republic of Ser- bia, within the project TR32035:“Development of Dialogue Systems for Serbian …

Natural Language Generation in the context of Multimodal Interaction in Portuguese JC Pereira, AJS Teixeira, JS Pinto – Electrónica e Telecomunicações, 2012 – revistas.ua.pt … It is mainly used in industrial dialogue systems. … between the two sentence- aligned files with corpus (GIZA++), and for building and applying statistical language models (SRILM). … SimpleNLG[35, 36] is a Java API toolkit, developed under the supervision of Ehud Reiter at the … Cited by 1 Related articles

A language modeling approach to identifying code-switched sentences and words LC Yu, WC He, WN Chien – CLP 2012, 2012 – aclweb.org … When dealing with code-switched input, intelligent systems such as dialog systems must be capable of identifying the various languages and recognize … N-gram models for both code-switching and non- code-switching were trained using the SRILM toolkit (Stolcke, 2002) with n … Cited by 1 Related articles All 9 versions

What if everyone could do it?: a framework for easier spoken dialog system design P Milhorat, S Schlögl, G Chollet, J Boudy – Proceedings of the 5th ACM …, 2013 – dl.acm.org … Stolcke, A. SRILM-An extensible language modeling toolkit. In Proc. of ICSLP (2002). 28. Williams, JD, and Young, S. Partially observable Markov decision processes for spoken dialog systems. Computer Speech & Language 21, 2 (2007), 393–422. 29. … Cited by 2 Related articles All 4 versions

A unified framework for translation and understanding allowing discriminative joint decoding for multilingual speech semantic interpretation B Jabaian, F Lefèvre, L Besacier – Computer Speech & Language, 2014 – Elsevier … The framework can be generalized to other components of a dialogue system. Abstract. … It has been shown that using lattices as SLU input decreased the classification error rate in the case of the AT&T’s “How May I Help You” dialogue system. In parallel Servan et al. … Cited by 1 Related articles All 5 versions

Neural network language models for off-line handwriting recognition F Zamora-Martínez, V Frinken, S España-Boquera… – Pattern Recognition, 2014 – Elsevier Unconstrained off-line continuous handwritten text recognition is a very challenging task which has been recently addressed by different promising techniques. T. Cited by 9 Related articles All 7 versions

Language style and domain adaptation for cross-language SLU porting E Stepanov, I Kashkarev, AO Bayer… – … (ASRU), 2013 IEEE …, 2013 – ieeexplore.ieee.org … The hidden-ngram tool from SRILM toolkit [9], which tags a sequence of to- kens with hidden events occurring between them, is used … In a live dialog system these entities are usually handled by their associated grammars, either handcrafted by the developers or provided as built … Cited by 3 Related articles All 3 versions

Automatic dialogue acts classification in Slovak dialogues M Pleva, S Ondas, J Juhar – … RADIOELEKTRONIKA), 2015 25th …, 2015 – ieeexplore.ieee.org … Whereas DM methods based on state machines, frames or plans are often used in task-oriented spoken or multimodal dialogue systems, their usage for conversational machines, where dialog interaction is more … Bigram models were prepared using the SRILM toolkit. …

Multimodal interaction for information retrieval using natural language A Revuelta-Martínez, L Rodríguez… – Computer Standards & …, 2013 – Elsevier … In this case, we discuss a spoken dialog system which is able to help the user when performing certain predefined actions within the database. … In Section 6 we explain how to extend the initial approach by adding a spoken dialog system in order to provide a better assistance. … Cited by 5 Related articles All 2 versions

Using Recurrent Neural Networks for Slot Filling in Spoken Language Understanding G Mesnil, Y Dauphin, K Yao, Y Bengio… – Audio, Speech, and …, 2015 – ieeexplore.ieee.org … observation models differently. We do so by introducing a tunable model combination weight, , whose value is optimized on held-out data. For computation, we used the SRILM toolkit (http://www.speech.sri.com/projects/srilm/). … Cited by 7 Related articles All 7 versions

Initial experiments with Tamil LVCSR J Melvin Jose, NT Vu, T Schultz – Asian Language Processing ( …, 2012 – ieeexplore.ieee.org … In [3] and [4], a Tamil spoken dialog system for farmers in Tamil Nadu and a syllable-based Tamil recognizer … First, we built a word-level Language Model (LM) using the SRI Language Modeling Toolkit [9]. Afterwards, we use the … SRILM – An extensible language modeling toolkit. … Cited by 1 Related articles All 6 versions

OpenCCG Realizer Manual M White – Documentation of the OpenCCG Realizer, 2012 – svn.kwarc.info … with the SRILM toolkit [Sto02], as described in Section 4.7; in principle, other toolkits could be … tables are efficiently stored in a trie data structure (as in the SRILM toolkit), thereby avoiding any … As discussed in [Whi04b], with dialogue systems like COMIC n-gram models can do an … Cited by 3 Related articles

Simple Gesture-based Error Correction Interface for Smartphone Speech Recognition Y Liang, K Iwano, K Shinoda – … Annual Conference of …, 2014 – mazsola.iit.uni-miskolc.hu … In order to generate the WCN based candidate list, we em- ployed the SRILM toolkit [28]. … Interspeech, 2010. [24] K. Yoshino, S. Mori, and T. Kawahara, “Incorporating Semantic Information to selecion of Web Texts for language model of spo- ken dialogue system”, in Proc. … Cited by 1 Related articles All 5 versions

Alternative hypothesis generation using a weighted kernel feature matrix for ASR substitution error correction CH Liu, CH Wu, D Sarwono – Chinese Spoken Language …, 2012 – ieeexplore.ieee.org … in the applications of speech and natural language processing, such as Computer-Assisted Language Learning (CALL) and Spoken Dialogue Systems (SDS), usually … The n-gram language model is trained using SRILM toolkit with the setup of bi-gram on Chinese words. Fig. … Related articles All 2 versions

Robust dialogue act detection based on partial sentence tree, derivation rule, and spectral clustering algorithm CP Chen, CH Wu, WB Liang – EURASIP Journal on Audio, Speech, and …, 2012 – Springer … Figure 1 Block diagram of a spoken dialogue system. At turn t, the user utters U, which is recognized by ASR to be W. ? is a semantic representation of user’s intended dialogue act. … The lexicon contains 297 words. The bi-gram language model is estimated by SRILM toolkit [40]. … Cited by 2 Related articles All 10 versions

A language model for highly inflective non-agglutinative languages S Ostrogonac, D Miskovic, M Secujski… – … (SISY), 2012 IEEE …, 2012 – ieeexplore.ieee.org … styles. The language modeling in the experiment was done using the SRILM toolkit [14]. Trigram concept was adopted for all models. … 180 – Page 5. Dialogue Systems in Serbian and other South Slavic Languages” (TR-32035). REFERENCES … Cited by 3 Related articles All 3 versions

Spoken language processing in a conversational system for child-robot interaction. I Kruijff-Korbayová, H Cuayáhuitl, B Kiefer, M Schröder… – WOCCI, 2012 – macs.hw.ac.uk … While most work in spoken dialogue system de- velopment is based on pipeline architectures, there are notable exceptions such as [5, 6], which … training tools for acoustic models, however any tool that creates acoustic mod- els in the Hidden Markov Model Toolkit (HTK) format … Cited by 13 Related articles All 6 versions

Ontology-based pattern generator and root semantic analyser for spoken dialogue systems Y Benahmed, S Selouani… – Electrical & Computer …, 2012 – ieeexplore.ieee.org … In this paper, we present an effective framework for generat- ing patterns from root semantical patterns for use with an interac- tive spoken dialog system belonging to the third category, using a … [16] A. Stolcke, “Srilm – an extensible language modeling toolkit,” 2002, pp … Related articles All 3 versions

Impact of training corpus size on the quality of different types of language models for Serbian S Ostrogonac, M Secujski… – … Forum (TELFOR), 2012 …, 2012 – ieeexplore.ieee.org … All of these models are based on the n-gram concept and have been trained using the SRILM toolkit [1]. Three types … was supported by the Ministry of Education, Science and Technological Development of Serbia within the Project “Development of Dialogue Systems in Serbian … Related articles All 2 versions

A Graph-based Cross-lingual Projection Approach for Spoken Language Understanding Portability to a New Language S Kim – 2013 – oar.a-star.edu.sg … Index Terms— Spoken Dialogue Systems, Spoken Lan- guage Understanding, Language Portability, Statistical Ma- chine Translation 1. INTRODUCTION … 45, p. 2. [15] A. Stolcke et al., “Srilm-an extensible language mod- eling toolkit,” in Proceedings of the … Cited by 1 Related articles All 2 versions

Clarification Question Generation for Speech Recognition Error Recovery Using Monolingual SMT D Yu – Advanced Materials Research, 2013 – Trans Tech Publ … A 3-gram language model is trained with the SRILM toolkit for all generation models by using all CQs of the training data and a Chinese spoken language corpus1 (401,772 utterances). … 2] Purver M., “CLARIE: Handling Clarification Requests in a Dialogue System,” Research on … Related articles All 3 versions

Synthesizing expressive speech from amateur audiobook recordings E Szekely, TG Csapó, B Toth, P Mihajlik… – … (SLT), 2012 IEEE, 2012 – ieeexplore.ieee.org … needs in applications such as speech generating devices (SGD) for non-speaking individ- uals, speech-to-speech translation applications and dialogue systems using intelligent … A word trigram model was trained using the SRILM Toolkit [8] ap- plying Good-Turing discounting. … Cited by 9 Related articles All 4 versions

ASR domain adaptation methods for low-resourced languages: application to Romanian language H Cucu, L Besacier, C Burileanu… – … ), 2012 Proceedings of …, 2012 – ieeexplore.ieee.org … [4] B. Jabaian, L. Besacier, F. Lefevre, “Combination of Stochastic Understanding and Machine Translation Systems for Language Portability of Dialogue Systems,” ICASSP 2011, Prague … [10] A. Stolcke, “SRILM – an extensible language modeling toolkit,” ICSLP 2002 … Cited by 6 Related articles All 8 versions

Improved recognition of Hungarian call center conversations B Tarjan, G Sárosi, T Fegyo… – Speech Technology and …, 2013 – ieeexplore.ieee.org … By default statistical n-gram models were estimated with SRI Language Modeling toolkit (SRILM) [20] by using modified Kneser-Ney pruning … for the recognition of non-verbal vocalisations in conversational speech,” in Perception in multimodal dialogue systems, Springer Berlin … Cited by 1 Related articles All 4 versions

Unsupervised data processing for classifier-based speech translator E Ettelaie, PG Georgiou, SS Narayanan – Computer Speech & Language, 2013 – Elsevier … 2006). It has been also used in a range of other applications such as systems with virtual interactive characters or machine spoken dialog systems to implement speech understanding (Leuski et al., 2006 and Traum et al., 2007). … Cited by 2 Related articles All 4 versions

Automated transcription of conversational Call Center speech–with respect to non-verbal acoustic events G Sárosi, B Tarján, T Fegyó… – Intelligent Decision …, 2014 – speechlab.tmit.bme.hu … 2k complexity. Consequently, by default for all the tasks word-based, trigram language models with Kneser-Ney smoothing [22] were built by using the SRI Language Modeling toolkit (SRILM) [23]. 5.1 Training text corpora As … Related articles All 4 versions

Large vocabulary Russian speech recognition using syntactico-statistical language modeling A Karpov, K Markov, I Kipyatkova, D Vazhenina… – Speech …, 2014 – Elsevier … used. In (Bechet and Nasr, 2009), a syntactic parser for spontaneous speech recognition outputs is used for identification of verbal sub-categorization frames for dialogue systems and spoken language understanding tasks. The … Cited by 19 Related articles All 6 versions

Generalization of discriminative approaches for speech language understanding in a multilingual context B Jabaian, F Lefèvre, L Besacier – Statistical Language and Speech …, 2013 – Springer … Jabaian, B., Besacier, L., Lef`evre, F.: Combination of stochastic understanding and machine translation systems for language portability of dialogue systems. In: ICASSP (2011) 13. … In: INTERSPEECH (2006) 30. Stolcke, A.: Srilm-an extensible language modeling toolkit. … Cited by 1 Related articles All 8 versions

Tokenizing fundamental frequency variation for Mandarin tone error detection R Tong, NF Chen, BP Lim, B Ma… – Acoustics, Speech and …, 2015 – ieeexplore.ieee.org … The GMM index sequences derived from the same tone class are used to train n-gram language model using SRILM toolkit. … Edlund, and Mattias Heldner, “An in stantaneous vector representation of delta pitch for speaker change prediction in conversation dialogue system,” in … Cited by 2

Investigating critical speech recognition errors in spoken short messages A Pappu, T Misu, R Gupta – Proceedings of IWSDS, 2014 – uni-ulm.de … ICASSP), Vancouver (Canada) (2013) 3. Bohus, D., Rudnicky, A.: Integrating multiple knowledge sources for utterance-level confi- dence annotation in the cmu communicator spoken dialog system. Tech. … Stolcke, A.: SRILM-an extensible language modeling toolkit. … Cited by 4 Related articles All 2 versions

SMT-based ASR domain adaptation methods for under-resourced languages: Application to Romanian H Cucu, A Buzo, L Besacier, C Burileanu – Speech Communication, 2014 – Elsevier … speakers. Several language adaptation methods for spoken dialog systems are proposed in [14] (English to Spanish) and [16] (French to Italian). … The implementation of the SMT system is based on the Moses Translation Toolkit [40]. Moses … Cited by 6 Related articles All 5 versions

Towards high-reliability speech translation in the medical domain G Neubig, S Sakti, T Toda, S Nakamura… – Proc. MedNLP, 2013 – phontron.com … In Proc. ACL, pages 157–166. Andreas Stolcke. 2002. SRILM – an extensible lan- guage modeling toolkit. In Proc. … 2000. Learning to predict problematic situations in a spoken dialogue system: experiments with how may I help you? In Proc. … Cited by 2 Related articles All 9 versions

Towards Empirical Dialog-State Modeling and its Use in Language Modeling. NG Ward, A Vega – INTERSPEECH, 2012 – cs.utep.edu … [4] A. Raux, N. Mehta, D. Ramachandran, and R. Gupta, “Dynamic language modeling using bayesian networks for spoken dialog systems,” in Interspeech, pp. 3030–3033, 2010. … [19] A. Stolcke, “SRILM – an extensible language modeling toolkit,” in Proc. Intl. Conf. … Cited by 8 Related articles All 5 versions

Topic identification techniques applied to dynamic language model adaptation for automatic speech recognition JD Echeverry-Correa, J Ferreiros-López… – Expert Systems with …, 2015 – Elsevier … based language models can be found in a broad spectrum of applications, such as in information retrieval systems as part of the ranking function (Zhai, 2008), in spoken dialogue systems for adapting … For this step, we have used the Freeling Toolkit ( Padró & Stanilovsky, 2012). … Related articles All 3 versions

Using crowdsourcing for grammar induction with application to spoken dialogue systems E Palogiannidi – 2013 – artemis.library.tuc.gr … 9 Page 11. Chapter 1 Introduction 1.1 Motivation Interactive Spoken Dialogue Systems, (SDS) are becoming increasingly pervasive in daily life. … perplexity := 2H(X) (2.3) For calculating perplexity in practice, we need a LM and a test corpus and we can use the SRILM toolkit [51] . … Related articles

Word informativity influences acoustic duration: Effects of contextual predictability on lexical representation S Seyfarth – Cognition, 2014 – Elsevier … a collection of English telephone conversations created at the Linguistic Data Consortium to aid speech-to-text dialogue systems (Cieri, Graff … were smoothed with the modified Kneser–Ney method described by Chen and Goodman (1998), using the SRILM Toolkit (Stolcke, 2002 … Cited by 5 Related articles All 7 versions

Recognizing Young Readers’ Spoken Questions. W Chen, J Mostow, Gregory Aist – IJ Artificial Intelligence in Education, 2013 – iaiedsoc.org … When little training data from real user is available, researchers in dialog systems have built language models from hand written … et al., 2003], and trained a bigram model on the resulting POS sequences using the SRILM language modeling toolkit (Stolcke, 2002). … Cited by 1 Related articles All 5 versions

Call routing based on a combination of the construction-integration model and latent semantic analysis: A full system G Jorge-Botana, R Olmos, A Barroso – Informatica, 2015 – informatica.si … Good-Turing. The package that we use to calculate probabilities in our model is SRILM [25]. It has Good-Turing as the default method (see http://www.speech.sri. com/projects/srilm/manpages/ngra m-discount.7.html). It works … Cited by 1 Related articles All 7 versions

A global model for concept-to-text generation I Konstas, M Lapata – Journal of Artificial Intelligence Research, 2013 – dl.acm.org … 37. Ratnaparkhi, A. (2002). Trainable approaches to surface natural language generation and their application to conversational dialog systems. Computer Speech ‘ Language, 16(3-4), 435- 455. … 43. Stolcke, A. (2002). SRILM – an extensible language modeling toolkit. … Cited by 5 Related articles

Statistical semantic interpretation modeling for spoken language understanding with enriched semantic features A Celikyilmaz, D Hakkani-Tur… – … Workshop (SLT), 2012 …, 2012 – ieeexplore.ieee.org … In a typical dialog system [1, 2, 3], a speech recognizer takes the user’s spoken utterances and converts them into text. … 219 Page 5. where p(g) is the prior probability for genre g. We trained LMgs using SRILM [31], with Kneser-Ney smooth- ing and using the default parameters. … Cited by 4 Related articles All 11 versions

A historical perspective of speech recognition X Huang, J Baker, R Reddy – Communications of the ACM, 2014 – dl.acm.org … Today, we can use open research tools, such as HTK, Sphinx, Kaldi, CMU LM toolkit, and SRILM to build a working system. … 42. Williams, J. and Young, S. Partially observable Markov decision processes for spoken dialog systems. … Cited by 19 Related articles All 3 versions

Design, development and field evaluation of a Spanish into sign language translation system R San-Segundo, JM Montero, R Córdoba… – Pattern Analysis and …, 2012 – Springer … This program is a beam search decoder for phrase-based statistical machine translation models. In order to obtain a 3-gram language model needed by Moses, the SRI language modelling toolkit has been used [ 35 ]. The translation based on SFST is made as set out in Fig. … Cited by 20 Related articles All 12 versions

Automatic Allophone Deriving for Korean Speech Recognition J Xu, Y Si, J Pan, Y Yan – Computational Intelligence and …, 2013 – ieeexplore.ieee.org … A trigram language model built by SRILM Toolkit is adopted. … Kashioka, Satoshi Nakamura, “Conditional Random Fields for Modeling Korean Pronunciation Variation,” Proceedings of the Paralinguistic Information and its Integration in Spoken Dialogue Systems Workshop 2011 … Related articles All 3 versions

Discriminative language modeling with linguistic and statistically derived features E Arisoy, M Saraçlar, B Roark… – Audio, Speech, and …, 2012 – ieeexplore.ieee.org … structure for ASR [5]; and semantic analysis for spoken dialogue systems [6]. A joint morphological- … See [45] for details of the boundary identification. In this paper we use the TextTiling algorithm implementation in Natural Language Toolkit (NLTK)3. … Cited by 16 Related articles All 4 versions

Patent Translation within the MOLTO Project Cristina España-Bonet Ramona Enache Adam Slaski Aarne Ranta L Màrquez, M Gonzàlez – Frontiers of Multilingual Grammar …, 2013 – 130.241.16.4 … Up to now, GF has been applied in several small-to-medium size domains such as dialogue systems 2 or the translation of mathematical exercises … Linguistic Issues in Language Technology 2 (1)(2009) [4] Stolcke, A.: SRILM–An extensible language modeling toolkit. … Related articles All 3 versions

Bilingual Continuous-Space Language Model Growing for Statistical Machine Translation R Wang, H Zhao, BL Lu, M Utiyama, E Sumita – 2013 – ieeexplore.ieee.org … The decoding speed of the grown LM is nearly the same as the normal n-gram LM, because the neural network probabilities are encoded into the grown n-gram LM using the SRILM toolkit. As a result, the proposed method has significant advantage on computational cost. … Related articles

A System for Recognizing Natural Spelling of English Words L Czech, A Waibel, S Stüker, DIT Köhler – Links, 2014 – isl.anthropomatik.kit.edu … There are plenty of use cases for such a system, including out-of-vocabulary learning, error recovery and dialog systems, for example in warehousing or flight booking. The recognizer consists of two main components which are based on the Janus recognition toolkit. … Related articles All 2 versions

Comparison of Methods for Topic Classification of Spoken Inquiries R Torres, H Kawanami, T Matsui, H Saruwatari… – Information and Media …, 2013 – jlc.jst.go.jp … AM training tool HTK 3.2 [18] Acoustic model PTM [19], 2,781 HMMs, 1,965 states, 8,256 mixtures Acoustic features 12 MFCC, 12 ? MFCC, ? E AM training Baum-Welch, 3 iterations LM training tool SRILM 1.5.0 [20] Language model 3-gram, Kneser-Ney smoothing LM … Related articles All 6 versions

Lattice generation with accurate word boundary in WFST framework Y Guo, Y Si, Y Liu, J Pan, Y Yan – Image and Signal Processing …, 2012 – ieeexplore.ieee.org … continuous speech recognition (LVCSR) has been applied in many areas including dictation systems, voice search, voice input systems, spoken term detection, spoken dialogue systems and so on. … [10] A. Stolcke, “Srilm-an extensible language modeling toolkit,” in … Cited by 2 Related articles

Leveraging Twitter for Low-Resource Conversational Speech Language Modeling A Jaech, M Ostendorf – arXiv preprint arXiv:1504.02490, 2015 – arxiv.org … 573–576. [9] K. Yoshino, S. Mori, and T. Kawahara, “Incorporating semantic infor- mation to selection of web texts for language model of spoken dialogue system,” in Acoustics … 1975–1978. [16] A. Stolcke, “SRILM – an extensible language modeling toolkit,” in Proc. … Related articles All 2 versions

Speech-to-speech translation to support medical interviews JASG Rodrigues – 2013 – repositorio.ul.pt Page 1. UNIVERSIDADE DE LISBOA Faculdade de Ciências Departamento de Informática SPEECH-TO-SPEECH TRANSLATION TO SUPPORT MEDICAL INTERVIEWS Jo˜ao Ant´onio Santos Gomes Rodrigues PROJETO MESTRADO EM ENGENHARIA INFORM´ATICA … Cited by 1 Related articles All 3 versions

An attribute detection based approach to automatic speech processing SM Siniscalchi, CH Lee – Loquens, 2014 – loquens.revistas.csic.es … This number, often referred to as a CM, serves as a reference guide for the dialogue system to provide an appropriate response to its users, just as an intelligent human being is expected to do when interacting with other people. … The SRILM (Stolcke, 2002) toolkit was used … Related articles

Automatic detection of speaker state: Lexical, prosodic, and phonetic approaches to level-of-interest and intoxication classification WY Wang, F Biadsy, A Rosenberg… – Computer Speech & …, 2013 – Elsevier … This focus is motivated primarily by Spoken Dialogue System (SDS) applications, such as call … trigram language models on the training set using the SRI Language Modeling Tookit (Stolcke … Indices) (Silverman et al., 1992) annotations using the AuToBI Toolkit (Rosenberg, 2010 … Cited by 4 Related articles All 6 versions

Automatic assessment of syntactic complexity for spontaneous speech scoring S Bhat, SY Yoon – Speech Communication, 2015 – Elsevier … Currently, speech-enabled dialog systems allow learners to practice their speaking and listening with a virtual interlocutor (eg, SpeakESL), to receive feedback on their pronunciation … We used the SRILM toolkit ( Stolcke, 2002) for training the models with Witten–Bell smoothing. … Related articles

Word Activation Forces-Based Language Modeling and Smoothing M Qin, G Liu, B Li, Y Lu – Intelligent Human-Machine Systems …, 2013 – ieeexplore.ieee.org … [1] F. Zamora, MJ Castro, and R. De-Mori, “Cache neural network language models based on long-distance dependencies for a spoken dialog system,” IEEE ICASSP, 2012, pp. … [6] S. Andreas, “SRILM an extensible language modeling toolkit,” In Proceedings of the … Related articles All 2 versions

Verification based ECG biometrics with cardiac irregular conditions using heartbeat level and segment level information fusion M Li, X Li – Acoustics, Speech and Signal Processing (ICASSP), …, 2014 – ieeexplore.ieee.org … 1–6. [5] A. Haag, S. Goronzy, P. Schaich, and J. Williams, “Emotion recogni- tion using bio-sensors: First steps towards an automatic system,” in Affective Dialogue Systems, pp. 36–48. … 1–7. [38] A. Stolcke, “Srilm-an extensible language modeling toolkit.,” in IN … Cited by 1 Related articles All 5 versions

Theoretical analysis of diversity in an ensemble of automatic speech recognition systems K Audhkhasi, AM Zavou, PG Georgiou… – Audio, Speech, and …, 2014 – ieeexplore.ieee.org … This result explains the trade-off between the WER of the individual systems and the diversity of the ensemble. We support this result through ROVER experiments using multiple ASR systems trained on standard data sets with the Kaldi toolkit. … Cited by 9 Related articles All 4 versions

Improving automatic classification of prosodic events by pairwise coupling C González-Ferreras… – Audio, Speech, and …, 2012 – ieeexplore.ieee.org Page 1. Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. … Cited by 7 Related articles All 5 versions

Portuguese text generation using factored language models EM de Novais, I Paraboni – Journal of the Brazilian Computer Society, 2013 – Springer … presently disregarded. The proposed models are summarised in Fig. 3. Unless specified otherwise, all models were built using SRILM [33] and using the default tool parameters and a back-off strategy as follows. The first model … Cited by 9 Related articles All 4 versions

A Graph-Based Approach to String Regeneration M Horvat, W Byrne – Student Research Workshop at EACL, 2014 – cl.cam.ac.uk … 3. The LKH heuristic TSP solver was created by Keld Helsgaun. 4. The language model server is a part of the SRILM toolkit created by the SRI Speech Technology and Research Laboratory. … tems, and Dialogue Systems, suffer from the same problem (Soricut and Marcu, 2005). … Cited by 2 Related articles All 12 versions

Discriminative reranking for spoken language understanding M Dinarelli, A Moschitti… – Audio, Speech, and …, 2012 – ieeexplore.ieee.org Page 1. Copyright (c) 2011 IEEE. Personal use is permitted. For any other purposes, permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. … Cited by 13 Related articles All 5 versions

Incremental Tree Substitution Grammar for Parsing and Sentence Prediction F Sangati, F Keller – Transactions of the Association for …, 2013 – tacl2013.cs.columbia.edu … Kamide, 1999). Also language processing systems often deal with speech as it is spoken, or text as it is being typed. A dialogue system should start interpreting a sentence while it is being spoken, and a question answering system … Cited by 3 Related articles All 10 versions

Combining multiple parallel streams for improved speech processing JTR de Sousa Miranda – 2014 – l2f.inesc-id.pt … retrieval engine to locate the most relevant documents. • Spoken dialog systems [56, 32] collaborate with a user in spoken language to complete a … In other situations, namely when the user is unfamiliar with the task at hand, or needs to have their hands free, dialog systems … Related articles

Spoken term detection ALBAYZIN 2014 evaluation: overview, systems, results, and discussion J Tejedor, DT Toledano, P Lopez-Otero… – EURASIP Journal on …, 2015 – Springer … for system training. The system is based on multi-lingual bottle-neck DNNs and Hidden Markov Model Toolkit (HTK) [ 83 ] for training and decoding and the IBM keyword search system for term detection [ 84 ]. Results showed …

Distributed speech translation technologies for multiparty multilingual communication S Sakti, M Paul, A Finch, X Hu, J Ni, N Kimura… – ACM Transactions on …, 2012 – dl.acm.org … We applied this technique to improve the quality of dialog systems by building and combining class-based models for interrogative and declarative sentences [Finch and Sumita 2008]. 3.3. Speech Synthesis Our current speech-translation system utilizes XIMERA [Kawai et al. … Cited by 1 Related articles

Enabling Non-Speech Experts to Develop Usable Speech-User Interfaces A Kumar – 2014 – reports-archive.adm.cs.cmu.edu … graphical interfaces, spoken dialog systems are being explored to provide access to relevant … optimizing a speech recognizer. The intended user of our toolkit called Speech Toolkit for Non- … thesis. 1.1. Current Approaches: Toolkits in Speech Recognition … Related articles All 3 versions

Detecting Off-Task Speech W Chen – 2012 – scs.cmu.edu … robot cooperation. Off-task speech input to computers presents both challenges and opportunities for such dialog systems. … As a prevalent speech event, off- task speech has been acknowledged in many dialog systems, such as health communication … Cited by 1 Related articles All 17 versions

Using emotion as inferred from prosody in language modeling SA Karkhedkar – 2013 – Citeseer … increasing. In particular, interactive spoken dialog systems have become commonplace with improvements in speech recognition technology. … devoted primarily towards (1) the dialog manager module of a spoken dialog system to decide … Related articles All 6 versions

Statistical post-editing and quality estimation for machine translation systems H Bechara – 2014 – doras.dcu.ie … Post-edited output is of higher quality but naturally more time-consuming and expensive than raw MT output. Several toolkits to facilitate human post-editing have been released along with machine translation tools and studies into … Cited by 2 Related articles All 2 versions

Adequacy–fluency metrics: Evaluating MT in the continuous space model framework RE Banchs, LF D’Haro, H Li – Audio, Speech, and Language …, 2015 – ieeexplore.ieee.org … each task and language. The models were computed with the SRILM toolbox [38]. As seen from (6), (9) and (10), different from other conventional metrics that compute matches between translation outputs and references, in … Cited by 1 Related articles

Natural Language Processing for Social Media A Farzindar, D Inkpen – Synthesis Lectures on Human …, 2015 – morganclaypool.com … Semantic Role Labeling Martha Palmer, Daniel Gildea, and Nianwen Xue 2010 Spoken Dialogue Systems Kristiina Jokinen and Michael McTear 2009 Introduction to Chinese Natural Language Processing Kam-Fai Wong, Wenjie Li, Ruifeng Xu, and Zheng-sheng Zhang 2009 …

[BOOK] Interactive Approaches to Video Lecture Assessment K Riedhammer – 2012 – books.google.com … 3.8 Language Modeling . . . . . 3.8.1 Statistical n-gram Models 3.8.2 Smoothing Techniques for Statistical n-gram Models 3.9 The KALDI Speech Recognition Toolkit . . . . . 3.10 Speech Recognition Performance Measures . . . . . … Cited by 3 Related articles All 4 versions

Finnish Language Speech Recognition for Dental Health Care S PUHEENTUNNISTUS – 2012 – users.marjaniemi.com Page 1. FINNISH LANGUAGE SPEECH RECOGNITION FOR DENTAL HEALTH CARE SUOMENKIELINEN PUHEENTUNNISTUS HAMMASHUOLLON SOVELLUKSISSA A THESIS SUBMITTED TO THE DEPARTMENT OF INFORMATION AND COMPUTER SCIENCE OF … Related articles All 4 versions

Improving statistical machine translation using bayesian word alignment and gibbs sampling C Mermer, M Saraçlar… – Audio, Speech, and …, 2013 – ieeexplore.ieee.org … ?e,f = ? = 0.0001 to obtain a sparse Dirichlet prior. After alignments were obtained in both translation directions, standard phrase-based SMT systems were trained in both directions using Moses [34], SRILM [35], and ZMERT [36] tools. The translations were … Cited by 4 Related articles All 8 versions

Semi-Supervised Semantic Tagging of Conversational Understanding using Markov Topic Regression. A Celikyilmaz, DZ Hakkani-Tür, G Tür, R Sarikaya – ACL (1), 2013 – aclweb.org … We built a language model using SRILM (Stol- cke, 2002) on the domain specific sources such as top wiki pages and blogs on online movie reviews, etc., to obtain the probabilities of domain-specific n-grams, up to 3-grams. … Cited by 1 Related articles All 14 versions

Translation rescoring through recurrent neural network language models Á PERIS ABRIL – 2014 – riunet.upv.es … such as automatic summarization, discourse analysis, machine translation, morphological segmentation, natural language generation and understanding, speech recognition, topic segmentation and recognition, information retrieval, dialog systems, question answering, etc. … Related articles All 2 versions

Sequence-based pronunciation variation modeling for spontaneous ASR using a noisy channel approach H HOFMANN, S Sakti, H Chiori… – … on Information and …, 2012 – search.ieice.org … It utilizes a 3-gram LM which can be tuned up onto a 5-gram LM. The statistical models of our sys- tem were trained with special toolkits for language model- ing [12] and word alignment [13]. The translation process is performed by a tool called CleopATRa [14]. 3.2 Data Corpora … Cited by 1 Related articles All 7 versions

Automatic speech recognition for mixed dialect utterances by mixing dialect language models N Hirayama, K Yoshino, K Itoyama… – Audio, Speech, and …, 2015 – ieeexplore.ieee.org … Thus, the outputs of our system are capable of being inputs of existing methods of NLU. It compares the words in the common language with reference sentences in the common language to place emphasis on utilizing ASR results for NLU in spoken dialogue systems. … Related articles All 2 versions

A Submodularity Framework for Data Subset Selection K Kirchhoff, J Bilmes, K Wei, Y Liu, A Mandal, C Bartels – 2013 – DTIC Document Page 1. AFRL-RH-WP-TR-2013-0108 A SUBMODULARITY FRAMEWORK FOR DATA SUBSET SELECTION Katrin Kirchhoff Jeff Bilmes Kai Wei Yuzong Liu Kai Wei Arindam Mandal Chris Bartels Department of Electrical Engineering … Cited by 1 Related articles

Cross-Domain and Cross-Language Porting of Shallow Parsing E Stepanov – 2014 – eprints-phd.biblio.unitn.it Page 1. PhD Dissertation International Doctorate School in Information and Communication Technologies DISI – University of Trento Cross-Domain and Cross-Language Porting of Shallow Parsing Evgeny A. Stepanov Advisor: Prof. Dr. Ing. Giuseppe Riccardi … Related articles

Effective Use of Cross-Domain Parsing in Automatic Speech Recognition and Error Detection MA Marin – 2015 – digital.lib.washington.edu … information, we attempt to detect their location and extent (within the ASR hypothesis), as well as the type, in order to handle them effectively during the subsequent clarification request made by the dialog system component. In particular we are interested in two types … Cited by 1 Related articles

Pre-Processing MRSes T Bruland – In Proceedings of the 10th International Conference on …, 2013 – aclweb.org … ACE can parse and generate using the compiled grammar. Our goal is to create a pipeline for the NorSource grammar and use it to create small question-answer systems or dialogue systems. The first step in the pipeline is the parsing process with ACE. … Cited by 1 Related articles All 9 versions

Distributional phrasal paraphrase generation for statistical machine translation Y Marton – ACM Transactions on Intelligent Systems and …, 2013 – dl.acm.org Page 1. 39 Distributional Phrasal Paraphrase Generation for Statistical Machine Translation YUVAL MARTON, University of Maryland, Columbia University, and IBM TJ Watson Research Center Paraphrase generation has been … Cited by 5 Related articles

[BOOK] Language and computers M Dickinson, C Brew, D Meurers – 2012 – books.google.com … 5 Classifying Documents 5.1 Automatic document classification 5.2 How computers “learn” 5.3 Features and evidence 5.4 Application: Spam filtering 5.5 Some types of document classifiers 5.6 From classification algorithms to context of use 6 Dialog Systems 6.1 Computers that … Cited by 1 Related articles

State of the art on semantic retrieval of AV content beyond text resources MF Moens, GJ Poulisse, MM VRT – 2012 – Citeseer Page 1. © TOSCA-MP consortium: all rights reserved page i State of the art on semantic retrieval of AV content beyond text resources Deliverable D3.1 TOSCA-MP identifier: TOSCAMP-D3.1-v1.0.docx Deliverable number: D3.1 … Cited by 1 Related articles All 2 versions

Computational Terminology: Exploring Bilingual and Monolingual Term Extraction J Foo – 2012 – diva-portal.org Page 1. Linköping Studies in Science and Technology. Thesis, No. 1523 Computational Terminology: Exploring Bilingual and Monolingual Term Extraction by Jody Foo Submitted to Linköping Institute of Technology at Linköping … Cited by 2 Related articles All 5 versions

Detecting grammatical errors with treebank-induced, probabilistic parsers J Wagner – 2012 – core.ac.uk … 93 4.9 APP/EPP ratios for using SRILM’s unigram language model for EPP and Charniak’s parser for APP . . . . … 103 4.15 Effect of smoothing on unigram language model probabilities: SRILM’s Gold-Turing smoothing vs. naive smoothing on a subset of 2,000 BNC sen- … Cited by 5 Related articles All 9 versions