Differences

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

--- draft [2009/07/14 16:12]
ptacek
+++ draft [2009/09/30 20:38]
ptacek
@@ Line 1: / Line 1: @@
-====== Description of Czech Companion November Prototype ======
+====== Architecture Description ======
-photopal domena, nahranej korpus, ze 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 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 using statistical machine translation in order to
-typy odpovedi a zpusob jejich implementace, rucne vyrobene dafy pro nasledujici topics: Person_Retired, Person_in_productive_age, Child, Husband, Wife, Wedding, Christmas, Handle_stalled_dialog
+to bridge the gap between dis-fluent spontaneous speech and a standard grammatical sentence.
-NLP server s tectomt, ASR/TTS/SR client, connected over network
-XXX JPta
-advances in Czech NLU (on reconstructed spoken data): 100vet(?) rucne anotovat pos, a-tree, t-tree, IE predicates, Named Entities, DA pro eval in-domain testy
+Results of part-of-speech tagging are passed on to 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.
-pos ? analyzovat, generovat a kontrolovat 'jen' kde je rozdil ve forme?
+The Named Entity Recognition module then marks personal names and geographical locations. C5 based Dialog Act classifier combines lexical and morphological features to assess the type of user utterance (such as question, acknowledgement, etc.) that is a useful clue for Dialog Manager decisions.
-===== Speech Reconstruction =====
+In addition, when generating the system response, the dialogue manager will pass through the NLG module the information about the appropriate communicative function tag (CF, see the CZ TTS module) along with the sentence that is to be generated. NLG is also used to generate paraphrases of user input sentences.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.
-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 =====
-features: XXX Mirek/Johanka
-performance indicator: accuracy
-===== Syntactic Parsing =====
+<html><br/><hr/><br/></html>
-features: induce dependencies and labels
-performance indicator: f-measure
-v tipu je natrenovat MacDonnalda na dialog datech, ten task je do M42, ze bysme
+The Czech Companion follows the original idea of Reminiscing about the User's Photos,
+taking advantage of the data collected in the first phase of the project (using a Wizard-of-Oz setting). The full recorded corpora was transcribed, a 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.
-===== Semantic Parsing =====
+The architecture is the same as in the English version, i.e. a set of modules communicating through the Inamode (TID) backbone. However, the set of modules is different, see Figure 1. Regarding the physical settings, the Czech version runs on two notebook computers connected by a local network. One serves as a Speech Client, running modules dealing with ASR, TTS and ECA; the other one as an NLP Server.
-features: meaning representation with semantic roles (69 labels), coordinations, argument structure, partial ellipsis resolution, pronominal anaphora resolution,
-performance indicator: f-measure
-===== Information Extraction =====
+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.
-features: template based identification of predicates
-covering predicates from  before-mentioned set of DAFs.
-performance indicator: accuracy
-===== Named Entities Recognition =====
+The Knowledge Base consists of objects (persons, events, photos) 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: 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: za tohle bych vydaval klasifikator, co rozhoduje ,jestli se rekne 'To je smutné/veselé'. Tem adjektivum rucne priradim negative/positive sentiment.
-performance indicator: f-measure
-===== 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 vybrane akce
-===== 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 ]

Institute of Formal and Applied Linguistics Wiki

Differences