Online search has empowered users with access to vast amounts of information. However, current online interfaces do not permit users to manipulate the hits on a search engine result page (SERP). This leads to the user adapting his/her own search style to suit the search engine instead of being the other way round. We present Critisearch, an online search interface that allows users to manipulate hits by upvoting, downvoting and sorting them such that they can be arranged in a user-defined order. Critisearch was originally developed for middle school children. However, we found an interesting dearth of studies on how graduate students conduct searches. In order to evaluate how the manipulation of hits can benefit users, we conducted a longitudinal study with 10 graduate students who used Critisearch and/or other search engine/s of their choice for conducting the scholarly search over a three week period. Results indicate that using Critisearch for hit manipulation enabled metacognitive scaffolding (plan, filter, sort information) on the search engine interface especially in exploratory search contexts. Critisearch seems to facilitate a conversation with the interface by enabling marking of hits. In addition, Critisearch also appears to promote reflection with the upvote/downvote capabilities for marking of hits available to the user. The hit manipulation and metacognitive scaffolding on the Critisearch interface encourages users to conduct their search tasks in a more breadth-first fashion as opposed to the commonly used depth-first search strategy. Using qualitative analysis, we discovered three distinct types of search tasks that users perform in a scholarly context namely, specific exploration, needle-in-a-haystack and general exploration. This analysis provides a starting point for a better understanding information needs of users in a scholarly context and a classification of search tasks thus, adding to the existing body of literature on nature of online search tasks. / M.S. / Online search engines have been pivotal in ushering the information age which contributes to the rapid dissemination of electronic information. Today, online search engines empower users to access this vast amount of information using queries on the search engine interface. Over the years, the search engine capabilities have been growing richer in terms of personalized user experiences. However, users have limited agency when it comes to personalizing or ordering the actual results that are displayed on the search engine result page. Search engines do not allow the user to reorder or provide feedback regarding the results for a particular search term or set of keywords. Therefore, most users adapt to the interface in order to get the results that they desire, instead of the other way around. Critisearch is the first step in the direction of making the user interface adapt to the user's needs. It allows users to order and rank the search results as to how relevant they are to the search terms entered. Graduate students employ search engines in a technical and advanced sphere of learning but the researchers found that there weren't as many studies about how graduate students conduct searches. In order to evaluate the efficacy of the interface, a study with 10 graduate students was conducted who used Critisearch to conduct academic searches over a period of three weeks. Results indicate that Critisearch allows users to have a more interactive experience with the search engine interface and promotes an alternative style of online search.

There is growing literature on search behavior and using search for prediction of market share or macroeconomic indicators. This research explores investors' stock search behaviors and investigates whether there are patterns in stock returns using those for return prediction. Stock search behaviors may reveal common interest among investors. In the first study, we use graph theory to find investment habitats (or search clusters) formed by users who search common set of stocks frequently. We study stock returns of stocks within the clusters and across the clusters to provide theoretical arguments that drive returns among search clusters. In the second study, we analyze return comovement and cross-predictability among economically related stocks searched frequently by investors. As search requires a considerable amount of cognitive resources of investors, they only search a few stocks and pay high attention to them. According to attention theory, the speed of information diffusion is associated with the level of attention. Quick information diffusion allows investors to receive relevant information immediately and take instantaneous trading action. This immediate action may lead to correlated return comovement. Slow information diffusion creates latency between the occurrence of an event and the action of investors. The slower response may lead to cross-predictability. Making use of the discrepancy in information diffusion, we implement a trading strategy to establish arbitrage opportunities among stocks due to difference in user attention. This research enriches the growing IS literature on information search by (1) identifying new investment habitats based on user search behaviors, (2) showing that varying degrees of co-attention and economic linkages may lead to different speed of information diffusion (3) developing a stock forecasting model based on real-time co-attention intensity of a group economically linked stocks and (4) embarking a new research area on search attention in stock market. The methods in handling complex search data may also contribute to big data research. / text

This master thesis analyses the methods used for duplicity document detection and possibilities of their integration with a web search engine. It offers an overview of commonly used methods, from which it chooses the method of approximation of the Jaccard similarity measure in combination with shingling. The chosen method is adapted for implementation in the Egothor web search engine environment. The aim of the thesis is to present this implementation, describe its features, and find the most suitable parameters for the detection to run in real time. An important feature of the described method is also the possibility to make dynamic changes over the collection of indexed documents.

This dissertation performs economic analysis of search advertising from a comprehensive picture of the competition facing advertisers---by incorporating the price competition to endogenously investigate advertisers' bidding incentive, and taking into account consumers' online search and the unique information structure associated with the search advertising format. It consists of three essays based on game-theoretic modeling. The first essay studies the oligopolistic price competition among advertisers placed in different advertising positions, considering distinctive features of consumers’ online search behaviors. We find an interesting local-competition pattern in which direct price competition occurs only between advertisers adjacent to each other. The second essay integrates the price competition into the bidding competition and investigates the endogenous bidding incentives of advertisers with different competitive strengths. Surprisingly, we find that an advertising position with a better exposure may not always be profitable for the advertisers with competitive advantage, even if it is cost free. We also show that the bidding outcome might not align with the relative competitive strength. The third essay further considers the effects of organic listing as a competing information source on the sponsored bidding competition and the outcome performances in search advertising. It provides answers to questions such as whether and why advertisers with sufficient exposure from the organic list may still be willing to bid for top sponsored positions, and how the existence of organic listing affects search engine’s revenue, consumer surplus, and social welfare. / text

This paper studies airline customers' online search and purchase behaviors. Two fundamental aspects of online behavior are examined: (1) the link between search behavior and buying behavior and (2) the evolution of inter-temporal search and purchase decisions of strategic buyers. In the first study, we examine online customers' dynamic conversion behaviors using clickstream data. A new model based on Markov chains that incorporates discrete choices and decision-timing is proposed to capture key search effects on consumer decisions as well as dynamics of browsing behavior both within and across visits. Empirical results show that within-site search activities lead to strong consumer engagement and thus increase purchase and revisit propensities. Fit comparison between first and second order Markov chains allows us to conclude that consumer decisions are primarily influenced by the current search. Furthermore, we observe that consumers dynamically adjust their browsing behavior both within and across visits. The second study investigates the evolution of inter-temporal search and purchase decisions of strategic buyers. Risk neutral buyers follow simple behavioral rules based on future and current prices and options available. We show that the trade-off between waiting and purchasing will become less and less favorable to waiting. Price elasticity should therefore drop as departure date approaches. With stationary price distributions, search and purchase efforts increase with proximity to the deadline. We extend the base model to allow for price evolution and demand uncertainty. We find that increases in mean price and price dispersion may attenuate increasing propensities for search and purchase. We demonstrate our models through a logit estimation on a unique data set from a major online travel agency.