[ Skip to the content ]

Institute of Formal and Applied Linguistics Wiki


[ Back to the navigation ]

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
draft [2009/07/14 16:09]
ptacek
draft [2009/09/30 20:54] (current)
ptacek
Line 1: Line 1:
-====== Description of Czech Companion November Prototype ======+====== Architecture Description ======
  
-photopal domenanahranej korpusze na to sou dafy (reusing SHEFF DM intergrated through Inamode Relayer (TID)) vhodny, moreover reusable for expected pomdp DM from UOX (reuse states, let pomdp's do the topology and handle the states transitions, post november work), +The Czech Companion follows the original idea of Reminiscing about the User's Photos, 
-typy odpovedi zpusob jejich implementace, rucne vyrobene dafy pro nasledujici topics: Person_Retired, Person_in_productive_age, Child, Husband, Wife, Wedding, Christmas, Handle_stalled_dialog +taking advantage of the data collected in the first phase of the project (using a Wizard-of-Oz setting). The full recorded corpora was transcribeda manual speech reconstruction was done on 92.6% of utterances((Manual speech reconstruction is still in progress.)) and a pilot dialog acts annotation was performed on a sample of 1000 sentences.
-NLP server s tectomt, ASR/TTS/SR client, connected over network +
-XXX JPta+
  
-advances in Czech NLU (on reconstructed spoken data): 100vet(?rucne anotovat posa-treet-treeIE predicatesNamed EntitiesDA pro eval in-domain testy +The architecture is the same as in the English version, i.e. a set of modules communicating through the Inamode (TIDbackbone. Howeverthe set of modules is differentsee Figure 1. Regarding the physical settingsthe Czech version runs on two notebook computers connected by a local network. One serves as a Speech Clientrunning modules dealing with ASRTTS and ECA; the other one as an NLP Server. 
-pos ? analyzovat, generovat a kontrolovat 'jen' kde je rozdil ve forme?+
  
 +The NLU pipeline, DM, and NLG modules at the NLP Server are implemented using a CU's own TectoMT platform that provides access to a single in-memory data representation through a common API. This eliminates the overhead of a repeated serialization and XML parsing that an Inamode based solution would impose otherwise.
  
-===== Speech Reconstruction ===== +<html><br/><hr/><br/></html>
-features: omit filler phrases, irrelevant speech events, false starts, repetitions, corrections, syntax smoothing (WO,  +
-imlementation(zahrnout tuhle info?): moses natrenovany na korpusu +
-performance indicator: BLEU score (overall scoring of all features) to annotated corpora from T5.2.1., nejaka baseline +
-XXX Mirek+
  
-===== Morphology Analyzer and POS tagging ===== +The ASR module based on Hidden-Markov models transforms input speech into text, providing a front-end between the user and the Czech demonstrator. The ASR output is smoothed into a form close to standard written text by the Speech Reconstruction module in order to bridge the gap between dis-fluent spontaneous speech and a standard grammatical sentence.
-features: XXX Mirek/Johanka +
-performance indicator: accuracy+
  
-===== Syntactic Parsing ===== +Results of the part-of-speech tagging are passed on to the Maximum Spanning Tree syntactic parsing module. A tectogrammatical representation of the utterance is constructed once the syntactic parse is available. Annotation of the meaning at tectogrammatical layer is more explicit than its syntactic parse and lends itself for information extraction. The Named Entity Recognition module then marks personal names and geographical locations. Afterwards, the dialog act classifier uses number of lexical and morphological features to assess the type of user utterance (such as questionacknowledgementetc.) that is a useful clue for Dialog Manager decisions.
-features: induce dependencies and labels +
-performance indicator: f-measure +
-v tipu je natrenovat MacDonnalda na dialog datechten task je do M42ze bysme +
  
 +The dialog is driven by a Dialog Manager component by USFD (originally developed for the English Senior Companion prototype).
 +CU has supplied the transition networks covering following topics: retired_person, husband, child, wife, wedding and Christmas.
 +Dialogue Manager provides information about the appropriate communicative function along with the sentence that is to be generated to the NLG module. The TTS module integrated with the TID avatar transforms system responses from the text form into the speech and visual (face expressions, gestures) representation. As such, it provides an interface between the demonstrator and the user.
  
-===== Semantic Parsing ===== +The Knowledge Base consists of objects (personseventsphotos) that model the information acquainted in the course of dialog. Those objects also provide a very basic reasoning (e.g. accounting for the link between date of birth and age properties). Each object's property is able to store multiple values with a varying level of confidence((Provided either by ASR module or from lexical clues contained in respective utterance.))and values restricted to a defined time span.
-features: meaning representation with semantic roles (69 labels)coordinationsargument structure, partial ellipsis resolution, pronominal anaphora resolution, +
-performance indicator: f-measure+
  
-===== Information Extraction ===== 
-features: template based identification of predicates 
-covering predicates from  before-mentioned set of DAFs. 
-performance indicator: accuracy 
- 
-===== Named Entities Recognition ===== 
-features: detect person names, geographical locations (organizations jsou potreba?) 
-performance indicator: f-measure 
- 
-===== Dialog Act Tagging ===== 
-features: tagset derived from DAMSL-SWBD, DA is a key feature driving 
-performance indicator: 
- 
-===== Sentiment Analysis ===== 
-features:  
-performance indicator: 
- 
- 
-===== Complete System Evaluation ===== 
-T5.2.7 tohle zminuje, nick webb to pro nas asi neudela 
-performance indicator: pocet slov ve vypovedich uzivatele(?),  
- 
- 
-===== Dialog Manager ===== 
-features: reply types, using (language independed) predicates (prakticky to znamena, ze pojmenuju testy na prechodech v dafech anglicky) 
-performance indicator: rucni hodnoceni prijatelnosti vypovedi 
- 
-===== Natural Language Generation ===== 
-features: variations, underspecified input (dott format), emotional markup (natvrdo v dafech a templatech u hodnoticich vet) 
-performance indicator: BLEU score 

[ Back to the navigation ] [ Back to the content ]