Dialogue and Information Retrieval Dialogs on Dialogs March

Intersections between Dialog Systems and IR n Current work n n Call Routing Question

Call Routing n n Task: given a NL expression of a problem, classify (route)

Call Routing (2) n n n It’s a classification problem! Salience (co-ocurrence) based approaches

The IR in Call Routing n n Regard the problem as text classification Do

The Dialog in Call Routing n Disambiguation n n Follow-up dialog n n Easy

Question Answering n n Task: answer to a question in Natural Language from a

IR in Question Answering n Everywhere: n n Document indexing Retrieval … What is

Dialog in QA n Refining the question: n n n Clarification dialogue Decide which

Why so little? n Different issues: n n n IR = lots of unstructured

But HUGE potential! n Voice-only random access to large amounts of information (“Voice IR”):

Slides: 11

Download presentation

Dialogue and Information Retrieval Dialogs on Dialogs March all the way through April 2003

Intersections between Dialog Systems and IR n Current work n n Call Routing Question Answering Why so little? What else? Let’s brainstorm!

Call Routing n n Task: given a NL expression of a problem, classify (route) it in one of several categories Examples n n n AT&T: How May I Help You British Telecom Jennifer Chu-Carroll

Call Routing (2) n n n It’s a classification problem! Salience (co-ocurrence) based approaches (AT&T) IR-like approaches (J. Chu-Carroll) n n Treat user requests as “documents” Use VSM and cosine similarity to classify

The IR in Call Routing n n Regard the problem as text classification Do standard IR work: n n LSA LDA Centroid vs. KNN approaches Results? Classification perf?

The Dialog in Call Routing n Disambiguation n n Follow-up dialog n n Easy to do based on the VSM IR approach HMIHY: frame-based follow-up dialogs Q: Is Call Routing dialog management? Q: Or is it more like understanding? Q: Why typical understanding/DM approaches fail in HMIHY-type domains?

Question Answering n n Task: answer to a question in Natural Language from a database of documents in Natural Language. Examples: n n http: //www. ai. mit. edu/projects/infolab/ http: //www. ask. com

IR in Question Answering n Everywhere: n n Document indexing Retrieval … What is different from traditional IR? n n Some parsing/understanding of questions and documents Some language generation (? )

Dialog in QA n Refining the question: n n n Clarification dialogue Decide which question to ask Only for very restricted domains uses fixed frames (Rutgers: HITIQA)

Why so little? n Different issues: n n n IR = lots of unstructured data, no NLP Dialog = structured data, lots of NLP Main problems: n n Structural mismatch NLU mismatch

But HUGE potential! n Voice-only random access to large amounts of information (“Voice IR”): n n n technical manuals of “in-the-field” devices (e. g. NASA) tutorial systems phone-based Google (e. g. legal information…) GUI+ for IR Learn dialog stuff from data (LM, NLG, parsing…)