FAO in OPAALS: Semantic Search Assistant

Margherita Sini, FAO 1 / 15

http://www.fao.org/aims/

FAO in OPAALS: WP6

OPAALS meeting

London, 10 December 2007

Margherita Sini, FAO 2 / 15

http://www.fao.org/aims/

WP6• Objectives

1. Metaphorology

2. Evolutionary framework of languages

3. Knowledge visualization

4. Common Ontology for epistemic cultures

• Tasks:– task 6.2: UniKassel objective 1

– task 6.4: LSE objective 2

– task 6.6: UniKassel, TUT, SUAS, LSE objective 2 & 4

– task 6.8: ITA, UniZar, UniKassel objective 4

– task 6.9: FAO, IITK objective 2 & 3

Margherita Sini, FAO 3 / 15

http://www.fao.org/aims/

Task 6.9• Title: Multilingual semantics in open knowledge

systems for agricultural community of practice• Deliverable D6.10: Multilingual Search Assistant

(MSA) (Due M34 = 1st April 2009)• Milestone M6.6: Development of a Multilingual

Search Assistant: challenges, experiences and the way to go (Due M30 = 1st Dec 2008)

• Partners: FAO and IITK– Andy Bagdanov, Soonho Kim, Margherita Sini, Johannes Keizer– Jeetendra Singh, Ajay Krishnan, Prabhakar T.V.

Margherita Sini, FAO 4 / 15

http://www.fao.org/aims/

Deliverable D6.10: MSA• Problems:

– current full text searches are based on frequency of text appearance, but this provides low precision and recall, since the tools don’t understand meaning;

– for an efficient search users need to know what information is in the collection to select proper keywords.

• Objectives: helping users to find relevant documents using their own language

Margherita Sini, FAO 5 / 15

http://www.fao.org/aims/

EN FR ES AR ZH

Cankers(EN) 100 1 0 0 100

Chancre(FR) 34 100 100 0 0

Necrosis cancerosa(ES) 2 0 100 0 0

(AR)قرح 0 0 0 100 0

癌肿 (ZH) 0 0 0 0 100

Margherita Sini, FAO 6 / 15

http://www.fao.org/aims/

MSA: Use case

navigateAGROVOC

build boolean expression

select multilingualand synonym

expansion

type term

user

select boolean

expression

select language

Margherita Sini, FAO 7 / 15

http://www.fao.org/aims/

MSA Component diagram

MSSAAPI

luceneindex

GUI

User

Results

Lucene-basedsearch engine

get query

set queryBrowsing

component

Booleancomponent

(build booleansearch)

Multilingualcomponent

(expand languagesand synonyms)

GUI

Margherita Sini, FAO 8 / 15

http://www.fao.org/aims/

System Workflow

end

start get query fromsearch engine

visualize AGROVOCbrowser

with the query

querynull?

visualize AGROVOCgraphical browser

(default value“agriculture”)

NOYESis

query stringin agrovoc?

YES

NO

user selectsa term

press ‘search”

user selects boolean navigator

user selectsmultilingual assistant

send query tosearch engine

visualize or updatequery string

on search textbox

Margherita Sini, FAO 9 / 15

http://www.fao.org/aims/

MSA Design

Margherita Sini, FAO 10 / 15

http://www.fao.org/aims/

MSA: 3 components

Margherita Sini, FAO 11 / 15

http://www.fao.org/aims/

Component n.1: Visual navigator

Margherita Sini, FAO 12 / 15

http://www.fao.org/aims/

Component n.2: Boolean query builder

Margherita Sini, FAO 13 / 15

http://www.fao.org/aims/

Component n. 3: multilingual and synonym expansion

DEMO

Margherita Sini, FAO 14 / 15

http://www.fao.org/aims/

Status• Preliminary investigations• Resources available:

– the AGROVOC Thesaurus– Web Services

• Some components available as prototypes• The system will be developed in Java

– integration of components needed– connection to lucene engine needed

• Possible extension with other resources (Wordnet, etc.), more semantics, etc.

Margherita Sini, FAO 15 / 15

http://www.fao.org/aims/

Thank you!Dankie

Faleminderit

Shukran

Merci

Rekhmet

Dziakuju

Dhanyabad

Nasima

Gràsce

Gràcies

Grazia

Dêkuji

Hvala

Tak

Dank u

Dankon

Tänan Tashakkur

Kiitos

Danke

Efcharisto

Toda Köszönöm

Takk

Agyamanac

Makasih

Gratias

Grazie

Arigato

Vandane

Webare

Asante

Kamsahamnida

Khawp jai Gratzia

Achiu

Aske

Terima kasih

Grazzi

Chjóonte

Shukriya

Cám ón

Dyakooyu

Tesekkurler Khawp khun