Natural language interface: desirable for a search syste

1
A Robust Ontology-Based Method for
TranslatingNatural Language Queries to
BK TP.HCM
HCMUT
Conceptual Graphs

• Tru Cao, Truong Cao, Thang Tran
• Presenter Hien Nguyen
• Semantic Web Group (VN-KIM)
• Faculty of Computer Science Engineering
• Ho Chi Minh City University of Technology

2
Outline

• Research motivation
• The roles of named entities
• Translation of queries to CGs
• Experimental results
• VN-KIM Search
• Concluding remarks

3
Research motivation

• Natural language interface desirable for a
  search system
• Machine translation for the general case
  saturated after many years of research
• What can make the difference limiting the domain
  of discourse to queries

4
The roles of named entities

• Ontological properties of NEs help identifying
  relations
• Query What county is Modesto, California in?

implicit relation determined by surrounding NEs
5
The roles of named entities

• NEs are used as anchors
• Query What county is Modesto, California in?

entities
6
The roles of named entities

• Relations among NEs are to be identified
• Query What county is Modesto, California in?

relations to be recognized
7
The roles of named entities

• An abstract view of a query
• Query
  

R3
E3
E2
E1
R1
R2
8
Translation of queries to CGs

• An ontology-based algorithm (9 steps)
• Recognizing specified entities
• Recognizing unspecified entities
• Extracting relational phrases
• Determining the type of queried entities
• Unifying identical entities
• Discovering implicit relations
• Determining the types of relations
• Removing improper relations
• Constructing the final conceptual graph

9
Translation of queries to CGs

• An ontology-based algorithm (steps 1-4)
• Recognizing specified entities
• What is the capital of Mongolia?
• Recognizing unspecified entities
• How many counties are in Indiana?
• Extracting relational phrases
• What state is Niagara Falls located in?
• Determining the type of queried entities
• What is WWE short for?

10
Translation of queries to CGs

• An ontology-based algorithm (steps 5-8)
• Unifying identical entities
• Who is the president of Bolivia?
• Discovering implicit relations
• What county is Modesto, California in?
• Determining the types of relations
• When was Microsoft established?
• Removing improper relations
• What city in Florida is Sea World in?

11
Translation of queries to CGs

• An ontology-based algorithm (step 9)
• Constructing the final conceptual graph
• What was the name of the movie that starred
  Sharon Stone and Arnold Schwarzenegger?

12
Translation of queries to CGs

• Advantages of conceptual graphs
• An intuitive graphical language
• Well-founded logical semantics
• Smooth mappping to and from natural language

13
Experimental results

• Tools and data
• KIM PROTON ontology 300 entity types and 100
  relation types (with enrichment)
• KIM World KB over 77,000 NEs (with enrichment)
• GATE NE recognition engine (with errors
  corrected)
• TREC 2007 with 445 queries

14
Experimental results

• Errors
• R-error due to GATEs performance
• O-error due to lack of entity types, relation
  types, NEs in KIM ontology and knowledge base
• Q-error due to expressiveness of simple
  conceptual graphs
• M-error due to the proposed algorithm itself

15
Experimental results
16
Experimental results
98
17
VN-KIM Search

• A semantic search engine

18
VN-KIM Search

• Searching by both named entities and keywords
• Clustering results by named entities and keywords
• Incorporating fuzzy matching
• Supporting natural language queries

19
Concluding remarks

• The proposed method is saving time off syntatic
  parsing and robust to grammatical mistakes

20
Concluding remarks

• Work under investigation
• Qualitative properties and non-binary relations
  are considered in ontology design
• Simple conceptual graphs are extended to
  represent queries with quantifiers, logical
  connectives, and comparative adjectives

21
Thanks for your attention

• VN-KIM Group
• http//www.cse.hcmut.edu.vn/vn-kim/