A Semantic Search Approach to Task-Completion Engines

Date: February 27, 2018
Venue: Stavanger, Norway. UiS TN910 - Innovation and Project Awareness

Please cite, link to or credit this presentation when using it or part of it in your work.

#InformationRetrieval #IR #TaskBasedSearch #TaskCompletionEngines #SemanticSearch

Transcript

1. A Semantic Search Approach to Task-Completion Engines Darío Garigliotti TN910 - Innovation and project awareness UiS - February 27, 2018
2. Background: Semantics - An entity is an individual or thing, uniquely identified Henrik Ibsen, Stavanger, UEFA Champions League - We describe entity properties using triples - Attributes: (Henrik Ibsen, birthdate, 20 March 1828) - Types: (Henrik Ibsen, is a, writer) - Relations: (Henrik Ibsen, work, A Doll’s House) - A knowledge base (KB) is a set of triples. - For us, semantics == structured knowledge 2
3. Semantic Search - "With great power comes great responsibility" - More users, greater expectations: understanding the search query - Search engines are becoming answer engines - Multiple techniques for query semantics - I am focusing on the next evolution stage: task completion 3
4. Objective - My research goal is to understand: - which challenges in semantic search are promissory for supporting task-completion engines, - which methods prove effective to model those challenges, - and how to integrate them into task-based search. 4
5. Techniques: Semantics in Current Search Engines (1) Type Other properties Entities and documents Entity 5
6. Semantics in Current Search Engines (2) 6
7. Status - I'm in my third year of studies in Computer Science - I have made contributions in these challenges: - utilizing entity type information for entity retrieval - understanding entity-oriented search intents - generating query suggestions to support task-based search 7
8. Stakeholders and Societal Factors - Social groups of relevance: Research 1. Colleagues in my research group Collaboration, integration 2. Researchers in the same area Contribution; competition 3. Researchers in related areas of the field Usefulness e.g. for evaluation; competition for areas 4. Researchers in related fields Usefulness of our/their developments 8
9. Stakeholders and Societal Factors - Social group of relevance: Industry 1. Commercial search engines Possible utilization of our developments 2. Major web-based services Possible utilization of our developments, benefit form our possible system demos reaching them 9
10. Stakeholders and Societal Factors - Social group of relevance: Developers - Benefit from our released codebase(s) - Social group of relevance: Users - Indirectly, when major engines or services implement our developments - Directly, only in minor scale through system demos 10
11. Stakeholders and Societal Factors - Interpretative flexibility - For users: a solution for completing tasks - For major commercial engines and services: to increase the quality and diversity of offers, then, their usage, and revenues - For other services: their promotion in responses - For researchers: contributions, and curiosity - Closure is hard to foresee 11
12. Challenges in RRI - Scalability - Whether our models and results are indeed extensible as working solutions on larger, real data - Data Privacy - Anonymization of datasets (collected by us or provided by previous work) - Development of tasks and evaluation methodologies oriented to protect privacy 12
13. RRI Measures - RRI goals - "respond to needs and ambitions of society, reflects its values, be responsible..." - "societal actors work together during the whole research process..." - privacy, security among the values by which to evaluate research outcomes and options 13
14. RRI Measures - EPSRC AREA framework - Reflect: ‣ the closer to a practical implementation of a working solution, the larger the reflections should be on motivations and implications: dilemmas, social transformations ‣ permanent reflection on assumptions, questions, areas of ignorance 14