Presentation is loading. Please wait.

Presentation is loading. Please wait.

Crosslingual Ontology-Based Document Retrieval (Search) in an eLearning Environment Eelco Mossel LSP 2007, Hamburg.

Similar presentations

Presentation on theme: "Crosslingual Ontology-Based Document Retrieval (Search) in an eLearning Environment Eelco Mossel LSP 2007, Hamburg."— Presentation transcript:

1 Crosslingual Ontology-Based Document Retrieval (Search) in an eLearning Environment Eelco Mossel LSP 2007, Hamburg

2 2 EU-Project LT4eL: Language Technology for eLearning ( ) Goal: use of Language Technology to improve the effectiveness of Learning Management Systems Multilingual Setting: 8 languages 12 European partner universities/institutes Crosslingual search: work together with: –Cristina Vertan (University of Hamburg) –Kiril Simov (Bulgarian Academy of Sciences, Sofia) –Alex Killing (ETH Zürich ( Eidgenössische Technische Hochschule) ) Framework

3 3 Project framework Learning Management System Goals of semantic search Resources for search function Features (user side) Internal components (developer side) Evaluation Overview

4 4 Learning Platform on the internet (interactive website) –Users can log in, have a profile, chat, … Fundamental: store and access Learning Material Units: Learning Objects (LOs) = documents Test-System : open source platform ILIAS ( Test material: text-based documents in 8 languages (PDF, HTML, MS-Word) Domain of test material: Computer Science for non CS specialists Learning Management System (LMS)

5 5 Goals of the approach 1. Improved retrieval of documents –Find documents that would not be found by simple text search (exact search word occurs in text) –Example: search for screen – retrieve doc that contains monitor but not screen. 2. Multilinguality –One implementation for all languages in the project 3. Crosslinguality –Find documents in languages different from search/interface language No need to translate search query Search possible with passive foreign language knowledge Crosslingual semantic search

6 6 Approach: use a domain ontology Creation: –Select keywords from LOs (also used for another project goal/task) –Choose keywords relevant for the domain Computer Science for Non-Computer Scientists) –Derive a set of concepts from the set of keywords. Concepts have English names/labels –Provide a definition in English for each concept –Create OWL taxonomy/ontology from concepts, by specifying relations between domain concepts, and mapping to DOLCE and WORDNET ontology. –1 ontology, for language-independent use, but contains English as common language for labels and definitions –Currently 707 concepts Resource: domain ontology

7 7 Connection between terms (words in a certain language) and concepts Create term-concept lexicons: –For each language and each concept, specify terms (synonyms if relevant) that denote the concept –At least one term for each concept –German: currently 939 terms 707 concepts Resource: term-concept lexicons

8 8 Concept Document relations Annotate concepts in documents in a semi-automatic way using the lexicons: –Occurrences of terms are annotated with corresponding concepts –Annotator (person) decides whether or not to annotate this occurrence, and chooses between concepts for ambiguous terms Resource: concept-annotations

9 9 Starting points A multilingual document collection An ontology including a domain ontology on the domain of the documents Concept lexicalisations in different languages Annotation of concepts in the documents Crosslingual semantic search

10 10 CommunicationsProtocol HTTPFTP Docs L1 Docs L3 Docs L2 HTTP SSL FTP IP Terms L1 Terms L3 Terms L2 Lexicons in different languages search terms browse and select Connecting the components

11 11 Search with ILIAS

12 12

13 13 Search using search words, concepts from the ontology or a combination of both –In case of combination: results of directly selected concepts come first Search for documents with super- or subconcepts –For documents in which desired concept is not found Ranking –Number of different searched concepts in document –Normalised annotation frequency –Super/subconcepts have lower weight Shared concepts (occurring in e.g. 50% of the found documents) –Example: Concept Report Some documents about academic writing Concept Publication Navigate through ontology (get related concepts) Features of the search functionality

14 14 Search functionality comprises: 1.Find terms in lexicons that reflect search query. 2.Find corresponding concepts for derived terms. 3.Find relevant documents for concepts. 4.Create ranking for set of found documents. 5.Create ontology fragment containing necessary information to present concept neighbourhood 6.Find shared concepts Internal components

15 15 Why start with a free text query? –User wants results fast (as in Google) –Compete with fulltext search and keyword search –Find starting point for ontology browsing Query lexicon: adopted/implemented strategies for –Tokenise create combinations for multiword terms (e.g. "space bar"), –Loose match of diacritic and uppercase letters (é e; E e) Other ideas to improve recognition of query: –Lemmatisation of search terms –Expansion of lexicon with word forms –Match similar strings Insertion of function words e.g. provedor acesso provedor de acesso –Automatic substring match -Dynamic list of available terms that contain input so far 1: Query Terms

16 16 Not always 1:1 mapping. Corresponding concept is missing from ontology –LT4eL: not in lexicon Unique result: term is lexicalisation of one concept Multiple concepts from one domain, e.g.: –Key (from keyboard) –Key (in database) Concepts from more domains: –Window (graphical representation on monitor) –Window (part of a building) Different concepts for different languages: –Kind (English: sort/type) –Kind (German: child) Let the user choose: present multiple browsing units 2: Term Concept

17 17 Simplest: –Disjunctive search with ranking For each concept, each document that is annotated with it is returned Documents with more desired concepts are ranked higher Use super/subconcepts Further possibilities –Conjunctive search: Combination of concepts must occur in a document Is taken into account in current ranking –Context search: Combination of concepts must occur in a paragraph or sentence –Word & Concept search combined: Document must contain concepts as well as certain words 3: Concept Documents

18 18 How useful is it, to find documents that treat a superconcept? –Negative example: lt4el:Subroutine lt4el:Software. Other children of Software are e.g.: Shareware, AuthoringLanguage –Positive example: GraphicalUserInterface UserInterface How useful is it, to find documents that treat a subconcept? –lt4el:Program has 93 subconcepts, e.g.: ApplicationProgram Computervirus Driver Unzip 3: Concept Documents (continued)

19 19 Does semantic search return correct results? (appropriate documents) How easy is it to use semantic search? Are the results better (precision/recall) than with keyword search or fulltext search (also available in ILIAS)? –Relevant for monolingual scenario Is the learning process improved? –Depends on quality of ontology and annotation –In multilingual case: depends on domain knowledge and language knowledge of multilingual test persons Evaluation

20 20 Thank you

Download ppt "Crosslingual Ontology-Based Document Retrieval (Search) in an eLearning Environment Eelco Mossel LSP 2007, Hamburg."

Similar presentations

Ads by Google