Constructing Complex Search Task with Subtasks to Improve Web Search and Sponsored Search Advertising

Abstract

Web users usually have a certain search goal before they submit a search query However many laypersons can’t transform their search goals into suitable queries Thus understanding original search goals behind a query is very important for search engines In the past decade many researches focus on classifying search goals behind a query into different search-goal categories including informational navigational and transactional In fact there may be more than one search goal behind a certain query In recent years web users often rely on search engines to deal with real-life complex tasks such as wedding planning travel planning or job hunting However conventional search engines intuitively consider a search query as a simple search task that can be accomplished by a single search goal In fact more and more queries are driven by complex tasks which include more than one subtask to be accomplished and thus force users to issue a series of subtask queries For example the complex task “prepare wedding” consists of multiple subtasks such as “buy wedding ring ” “reserve wedding venue ” and “rent wedding limousine ” Therefore to accomplish this complex task the users need to submit at least three subtask queries Furthermore each subtask in a complex task may trigger a few consumption needs To deal with subtasks consumption needs users may try to seek sponsored ads when the web search results fail to fit their consumption needs Guo and Agichtein (2010) mentioned that users searching on the web with consumption needs are more likely to click ads Hillard et al (2010) also pointed out that suggesting ideal ads may actually improve user search experience However given a complex task query conventional search engines are usually not able to provide suitable ads covering the consumption needs derived from the subtasks of a complex search task To enhance the advertiser’s revenue sponsored search systems should optimize the conversion rate of recommended ads Although some recent studies focus only on increasing the ad-click rate based on the click history data however improving ad-click rate may not accurately reflect if the clicked ads truly trigger users’ consumption needs and may not increase the conversion rate In fact some users probably click an ad due to curiosity and will not engage in any consumption in the advertised website When a user decides to purchase or reserve something the user probably achieved a consumption need Therefore suggesting attractive ads which fit the user’s true desires will also improve the conversion rate and decrease the search time for users Furthermore for advertisers a larger conversion rate with smaller click rate will optimize revenue and create a win-win situation To the best of our knowledge our work is the first to address the issue of complex task advertising while the above previous works only focus on single task advertising To help users to efficiently accomplish their complex search task in this work we first propose a search goal model based on the users’ search query Then we further propose a topic-event-based complex search task model to construct the structure of complex search tasks containing a complex task name and subtask names Finally based on our constructed complex task structures we propose a consumption-need-based complex-task ad recommendation model In this research work we developed a prototype system task-aware integrated search engine to help users efficiently accomplish their complex search tasks

Date of Award	2015 Aug 21
Original language	English
Supervisor	Wen-Hsiang Lu (Supervisor)