Intelligent Assisting Conversational Agents g g g viewed through - PowerPoint PPT Presentation

Intelligent Assisting Conversational Agents g g g viewed through novice users’ requests Mao Xuetao, Jean ‐ Paul Sansonnet, François Bouchet , , ç LIMSI ‐ CNRS Université Paris ‐ Sud XI IHCI 2009 (MCCSIS) Algarve, Portugal g g June 20th 2009

Outline � Problem � Assisting agents ss st g age ts � Assisting agents for web applications and services � The genealogy of the DIVA toolkit � A typical chatbot architecture � Advantages and drawbacks of the chatbot approach Advantages and drawbacks of the chatbot approach � Methodology � Methodology: a corpus ‐ based NLP ‐ chain � Methodology for the corpus collection � Excerpt from the sub ‐ corpus ‘Marco’ � Assistance is a linguistic genre � Implementation � DIVA NLP ‐ chain � DIVA semantic keys � DIVA formalization phase: ℜ ‐ rules ℜ DIVA f li ti h l � DIVA topic files � DIVA interpretation phase: ℑ ‐ rules � Conclusion Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 2

Outline � Problem � Assisting agents ss st g age ts Can we use the chatbot architectures as a base C h h b hi b � Assisting agents for web applications and services for the analysis and resolution of natural � The genealogy of the DIVA toolkit language assisting requests in web applications � A typical chatbot architecture and se vices? and services? � Advantages and drawbacks of the chatbot approach Advantages and drawbacks of the chatbot approach � Methodology � Methodology: a corpus ‐ based NLP ‐ chain � Methodology for the corpus collection � Excerpt from the sub ‐ corpus ‘Marco’ � Assistance is a linguistic genre � Implementation � DIVA NLP ‐ chain � DIVA semantic keys � DIVA formalization phase: ℜ ‐ rules ℜ DIVA f li ti h l � DIVA topic files � DIVA interpretation phase: ℑ ‐ rules � Conclusion Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 3

Assisting agents « An Assisting Agent is a software tool with the capacity to resolve help requests, issuing from novice users, about the static structure and the dynamic functioning of software components or services » (Maes, 1994) User User person with poor knowledge about the component (novice) person with poor knowledge about the component (novice) Request help demand in natural language (speech/text) rational, assistant, conversational, (can be embodied) Agent symbolic model of the structure and the functioning Mediator computer application, web service, ambient appliance t li ti b i bi t li C Component t Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 4

Assisting agents for web applications & services Keys issues: How can we improve • The precision y p of the Function of Assistance? • The genericity Th i it Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 5

The genealogy of the DIVA toolkit DOM CHS ECA Chatbots Contextual Help Embodied Natural Language Web Applications Systems Systems Conversational Agents Conversational Agents Processing Processing and services and services « Motivation Paradox » « Persona Effect » « Eliza Effect » Application model construction (Carroll & Rosson, 1987) (Weizenbaum, 1966) (Lester et al., 1997) (Leray & Sansonnet, 2005) Assistance Assistance Personification Personification Dialogue Dialogue Ubiquity Ubiquity ACA Webbots Assisting g Ludo social Ludo ‐ social Conversational Agents conversations Oral expression of frustration (Capobianco & Carbonell, 2002) Problem: How can we improve the DIVA precision and the genericity of the D D OM ‐ Integrated Virtual Agents I t t d Vi t l A t NLP ‐ chain in a web environment? Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 6

A typical webbot architecture Single pass, rule based, filtering process Single pass, rule based, filtering process “What is the XYZ “How to quit?” “How to do that ?” button for?” no no no Filtering by g y User natural U t l Filtering by the Filtering by the Evasive Evasive Recall of a R ll f the generic language utterance specific layer List preceding topic layer yes yes yes yes Specific answer linked to the Generic Recall of a Evasive character of the agent answer previous topic answer or to the task of the application h k f h li i Dialogical Common ‐ sense Minimalistic Task handling: g h handling dli di l dialogue session i Robustness precise but not generic handling Generic Customized Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 7

A typical finalized dialogue system Symbolic Customized User Application model of the Generic Modeling Lexicon application Reasoning tools about the Natural structure and the “User natural Formal Multimodal Language g g f functioning of the symbolic ti i f th b li language utterance” Semantic Request Reaction models of the application Analyzer and the users (tasks and plans handling) “I think that you speak too loud” ACTION: lower sound level “You speak too loud!” You speak too loud! SAY: “I speak lower now” JUDGE user [TOOMUCH(system.sound.level)] “You make too much noise, my dear” “The sound level is very very high” USER: update preferences “You make my ears ache” Etc. Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 8

Advantages and drawbacks of the chatbot approach � Advantages: easy, light, precise � They are easy to develop: no large semantic analyzer no complex reasoning tools; � They are easy to develop: no large semantic analyzer, no complex reasoning tools; � They are light to deploy in a web ‐ based environment � client architectures can be envisioned; � They provide robust natural language reactions (Evasive list effect – ELIZA effect); � They are tailored and well ‐ suited for the field of ludo ‐ social chat ; Th t il d d ll it d f th fi ld f l d i l h t � When associated with a given application, they can be customized to be extremely precise. � Drawbacks: lack of genericity D b k l k f i i � Minimalistic/ ultra ‐ customized model of the application; � Minimalistic model of the dialogue session and of the users; � No semantic analyzer � lack of precision in the requests (grammar, speech acts, …); � No formal requests � class reactions linked and dependant to specific linguistics patterns ; � No generic reasoning tools, especially when the function of assistance is concerned. � Need recoding quite everything for each new application, � No reusability Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 9

Outline � Problem Can we use the chatbot architectures as a base � Assisting agents ss st g age ts f for the analysis and resolution of natural th l i d l ti f t l � Assisting agents for web applications and services language assisting requests in web applications � The genealogy of the DIVA toolkit and services? � A typical chatbot architecture ― Yes, provided we improve drastically their , p p y � Advantages and drawbacks of the chatbot approach Advantages and drawbacks of the chatbot approach precision and genericity. � Methodology � Methodology: a corpus ‐ based NLP ‐ chain � Methodology for the corpus collection � Excerpt from the sub ‐ corpus ‘Marco’ � Assistance is a linguistic genre � Implementation � DIVA NLP ‐ chain � DIVA semantic keys � DIVA formalization phase: ℜ ‐ rules ℜ DIVA f li ti h l � DIVA topic files � DIVA interpretation phase: ℑ ‐ rules � Conclusion Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 10

Outline � Problem � Assisting agents ss st g age ts � Assisting agents for web applications and services � The genealogy of the DIVA toolkit � A typical chatbot architecture � Advantages and drawbacks of the chatbot approach Advantages and drawbacks of the chatbot approach � Methodology Because the linguistic domain of the Function of � Methodology: a corpus ‐ based NLP ‐ chain Assistance is precise and concise, we can rely on � Methodology for the corpus collection a corpus-based approach to exhibit the inherent � Excerpt from the sub ‐ corpus ‘Marco’ � Assistance is a linguistic genre generic phenomena. � Implementation � DIVA NLP ‐ chain � DIVA semantic keys � DIVA formalization phase: ℜ ‐ rules ℜ DIVA f li ti h l � DIVA topic files � DIVA interpretation phase: ℑ ‐ rules � Conclusion Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 11

Methodology: a corpus ‐ based NLP ‐ chain Collection of a corpus Collection of a corpus of natural language requests in assisting situations Base the genericity of the NLP-chain on Lexical Semantic Pragmatic phenomena occurring within the corpus classes keys classes Natural language Pragmatic handling Formal Multimodal syntactico-semantic “User utterance” User utterance heuristics Request Request R Reaction i analyzer l [Chatbot layers] Form [Chatbot layers] Mixed NLP-chain: � Dialogue systems: Intermediate formal form � Chat bots: rule-layers for each phase � Chat bots: rule layers for each phase “If I want to buy such a < QUEST IF THEUSER TOWANT TOOBTAIN such a Scenic, what can I do?” $THECAR WHAT TOCAN THEUSER TODO $THECAR WHAT TOCAN THEUSER TODO > < HOW TOOBTAIN $THECAR> Xuetao, Sansonnet, Bouchet ― LIMSI ‐ CNRS 12