Web search engines and complex information needs
Kuupäev
2012-09-07
Autorid
Ajakirja pealkiri
Ajakirja ISSN
Köite pealkiri
Kirjastaja
Abstrakt
Veebi otsingumootorid on muutunud põhiliseks teabe hankimise vahenditeks internetist. Koos otsingumootorite kasvava populaarsusega on nende kasutusala kasvanud lihtsailt päringuilt vajaduseni küllaltki keeruka informatsiooni otsingu järele. Samas on ka akadeemiline huvi otsingu vastu hakanud liikuma lihtpäringute analüüsilt märksa keerukamate tegevuste suunas, mis hõlmavad ka pikemaid ajaraame. Praegused otsinguvahendid ei toeta selliseid tegevusi niivõrd hästi nagu lihtpäringute juhtu. Eriti kehtib see toe osas koondada mitme päringu tulemusi kokku sünteesides erinevate lihtotsingute tulemusi ühte uude dokumenti. Selline lähenemine on alles algfaasis ja ning motiveerib uurijaid arendama vastavaid vahendeid toetamaks taolisi informatsiooniotsingu ülesandeid.
Käesolevas dissertatsioonis esitatakse rida uurimistulemusi eesmärgiga muuta keeruliste otsingute tuge paremaks kasutades tänapäevaseid otsingumootoreid. Alameesmärkideks olid:
(a) arendada välja keeruliste otsingute mudel,
(b) mõõdikute loomine kompleksotsingute mudelile,
(c) eristada kompleksotsingu ülesandeid lihtotsingutest ning teha kindlaks, kas neid on võimalik mõõta leides ühtlasi lihtsaid mõõdikuid kirjeldamaks nende keerukust,
(d) analüüsida, kui erinevalt kasutajad käituvad sooritades keerukaid otsinguülesandeid kasutades veebi otsingumootoreid,
(e) uurida korrelatsiooni inimeste tava-veebikasutustavade ja nende otsingutulemuslikkuse vahel,
(f) kuidas inimestel läheb eelhinnates otsinguülesande raskusastet ja vajaminevat jõupingutust ning
(g) milline on soo ja vanuse mõju otsingu tulemuslikkusele.
Keeruka veebiotsingu ülesanded jaotatakse edukalt kolmeastmeliseks protsessiks. Esitatakse sellise protsessi mudel; seda protsessi on ühtlasi võimalik ka mõõta. Edasi näidatakse kompleksotsingu loomupäraseid omadusi, mis teevad selle eristatavaks lihtsamatest juhtudest ning näidatakse ära katsemeetod sooritamaks kompleksotsingu kasutaja-uuringuid. Demonstreeritakse põhilisi samme raamistiku “Search-Logger” (eelmainitud metodoloogia tehnilise teostuse) rakendamisel kasutaja-uuringutes. Esitatakse sellisel viisil teostatud uuringute tulemused. Lõpuks esitatakse ATMS meetodi realisatsioon ja rakendamine parandamaks kompleksotsingu vajaduste tuge kaasaegsetes otsingumootorites.
Search engines have become the means for searching information on the Internet. Along with the increasing popularity of these search tools, the areas of their application have grown from simple look-up to rather complex information needs. Also the academic interest in search has started to shift from analyzing simple query and response patterns to examining more sophisticated activities covering longer time spans. Current search tools do not support those activities as well as they do in the case of simple look-up tasks. Especially the support for aggregating search results from multiple search-queries, taking into account discoveries made and synthesizing them into a newly compiled document is only at the beginning and motivates researchers to develop new tools for supporting those information seeking tasks. In this dissertation I present the results of empirical research with the focus on evaluating search engines and developing a theoretical model of the complex search process that can be used to better support this special kind of search with existing search tools. It is not the goal of the thesis to implement a new search technology. Therefore performance benchmarks against established systems such as question answering systems are not part of this thesis. I present a model that decomposes complex Web search tasks into a measurable, three-step process. I show the innate characteristics of complex search tasks that make them distinguishable from their less complex counterparts and showcase an experimentation method to carry out complex search related user studies. I demonstrate the main steps taken during the development and implementation of the Search-Logger study framework (the technical manifestation of the aforementioned method) to carry our search user studies. I present the results of user studies carried out with this approach. Finally I present development and application of the ATMS (awareness-task-monitor-share) model to improve the support for complex search needs in current Web search engines.
Search engines have become the means for searching information on the Internet. Along with the increasing popularity of these search tools, the areas of their application have grown from simple look-up to rather complex information needs. Also the academic interest in search has started to shift from analyzing simple query and response patterns to examining more sophisticated activities covering longer time spans. Current search tools do not support those activities as well as they do in the case of simple look-up tasks. Especially the support for aggregating search results from multiple search-queries, taking into account discoveries made and synthesizing them into a newly compiled document is only at the beginning and motivates researchers to develop new tools for supporting those information seeking tasks. In this dissertation I present the results of empirical research with the focus on evaluating search engines and developing a theoretical model of the complex search process that can be used to better support this special kind of search with existing search tools. It is not the goal of the thesis to implement a new search technology. Therefore performance benchmarks against established systems such as question answering systems are not part of this thesis. I present a model that decomposes complex Web search tasks into a measurable, three-step process. I show the innate characteristics of complex search tasks that make them distinguishable from their less complex counterparts and showcase an experimentation method to carry out complex search related user studies. I demonstrate the main steps taken during the development and implementation of the Search-Logger study framework (the technical manifestation of the aforementioned method) to carry our search user studies. I present the results of user studies carried out with this approach. Finally I present development and application of the ATMS (awareness-task-monitor-share) model to improve the support for complex search needs in current Web search engines.
Kirjeldus
Väitekirja elektrooniline versioon ei sisalda publikatsioone.
Märksõnad
Internet, veeb, infootsing, otsingumootorid, web, information retrieval, search engines