28 results
5 - Helping People Search
- from Part II - System SupportHelping People Search
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 141-200
-
- Chapter
- Export citation
-
Summary
The previous chapters have targeted actions that are useful as signals in models of relevance and satisfaction, as well as models of information seeking and use that have been developed to better understand the search process. One of the largest changes in information systems over the past few decades has been the transformation of search systems from tools that are used only by trained experts (such as reference librarians) to applications and services for the general population. In the early 1980s, researchers argued that information retrieval may become an elite activity unless search interfaces became easier for novices to use (Ingwersen, 1984). Supporting searching by the general populace involves both the simplification of query construction and result examination, as well as the introduction of tools to help people build queries and consider the results returned. In this chapter, I turn my attention to the mechanisms that have been developed to help people search. I focus on the search experience and searcher-facing components in particular, rather than the sophisticated methods employed during crawling, indexing, and ranking. Ranking algorithms and other backend components have been covered at length by other scholars (e.g., Manning et al., 2008; Croft et al., 2010); however, much still remains to be learned about the complexity of various interactive processes in search, and particularly about the effects of the variables involved (Belkin and Croft, 1992). Others provide excellent summaries of the history of search interfaces; both up to the end of past century (Marchionini and Komlodi, 1998; Shneiderman et al., 1999), and more recently (Hearst, 2009).
Emerging interest in areas such as human-computer information retrieval (HCIR) (Marchionini, 2006b) has directed attention toward the need to support users engaged in complex tasks and support more than simply information finding tasks. Information finding is handled well by existing search technologies. One of the central tenets of the design of search systems has been the need for searchers to assume control and responsibility over the search process. Search systems can offer assistance to searchers in performing their search tasks, but to benefit most from the search process, searchers must be engaged in the process and be responsible for their actions and the outcomes of the search.
References
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 415-498
-
- Chapter
- Export citation
14 - Call to Action
- from Part IV - Opportunities and Challenges
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 405-410
-
- Chapter
- Export citation
-
Summary
Opportunities and Challenges
The future is bright for search interaction. This chapter outlines some of the grand challenges for search interaction and present a call to action for the researchers and practitioners of search systems to work collaboratively to meet these challenges. The confluence of a number of important technological advances means that there is significant opportunity for the search community to advance the process of information seeking well beyond its current state. New interaction techniques, cloud-based storage, and mobile devices are some of the most important recent advances that need to be investigated more thoroughly by the research community when designing next-generation search systems. Some of the other notable recent advances, such as big data and machine learning, have led to significant advances in the development of support for the search process (Liu, 2009). The goals of search, both in terms of the core capabilities of search systems and searcher expectations for what these systems can do, need to be expanded beyond information finding (a facility that all search systems must possess) to promote the development of search technology to help people explore, learn, gain insights, and apply their knowledge. The remainder of this chapter outlines key opportunities presently available to the search community to shape the future of search interaction. These can be grouped into four categories: (1) experiences; (2) data; (3) evaluation; and (4) external engagement.
Development of search experiences
• Capitalize on new technologies and interaction paradigms. The pace of technological innovation is increasing rapidly. New devices are emerging that understand natural interactions such as touch, gesture, and voice, and can be used in many different settings. Although there has been some recent progress toward supporting the application of these new technologies in search interaction (see Hearst [2011] for a summary of some recent advances), research in the area of natural user information retrieval (NUIR) is still in its infancy. Despite the importance and growing prevalence of devices and applications with new capabilities, the research community is largely still fixated on supporting desktop-based Web searching, with the keyboard and the mouse as the only input mechanisms, and textual query statements and a ranked list of search results as the primary means of engaging with search systems. Although interest in search across different types of devices is growing (Montañez et al., 2014), little attention has been paid to new modalities.
1 - Introduction
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 1-12
-
- Chapter
- Export citation
-
Summary
Information seeking is a core life activity. People seek information for a number of reasons, including to find facts (Marchionini and Shneiderman, 1988) and to learn and facilitate effective decision making (Marchionini, 2006). Searchers may also have other motivations, such as pleasure and enjoyment (Wilson and Elsweiler, 2010). Historically, people have sought information through interactions with others, either synchronously through in-person or telephone dialog or asynchronously through written communication. Increasingly, information seeking is being conducted via automated search systems such as Web search engines. Meeting searcher requirements across the full spectrum of search goals that individuals may have is challenging with such generic search systems. Nonetheless, search engines are offering an increasing range of reactive and proactive search services to meet and anticipate searcher needs. Helping searchers with specific search tasks requires targeted search support, as well as different methods and criteria under which to evaluate the performance of search systems in different circumstances. For example, fact-finding tasks may require only a single resource or direct answer, and strong system performance may be evidenced by low searcher engagement and short task completion times. In contrast, when the goals involve exploration, learning, or enjoyment, richer search and exploration support may be needed (e.g., diverse results or query suggestions, dynamic information visualizations), and increased engagement over a much longer timespan may be a positive indication of system performance. Understanding the nature of the search task(s) being attempted is important in both designing and evaluating the support offered by search systems.
Effective interactions with automated search systems are a critical aspect of successful searching. These interactions can range in complexity from basic text query entry and result selection to rich gestural and spoken dialog interactions. Search tasks can also continue longitudinally and increasingly, these tasks transcend devices, domains, and applications. The range of search interactions, and the support that systems offer for performing them, is expanding given technological advances. Search systems are becoming more intelligent and more aware of their users’ interests and intentions, as well as of their surroundings when performing searches. This enables these systems to anticipate people's needs more accurately and to work symbiotically with searchers to support task completion directly.
Part III - Evaluation
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 305-306
-
- Chapter
- Export citation
-
Summary
I now turn to how interactive search systems should be evaluated, including the measures that should be employed, the methodologies, and the data and tools that are needed to perform these evaluations. This is particularly important if systems are being compared across multiple experimental sites and systems are being tested at scale with millions of searchers. Web search providers need to understand the performance of their systems at scale, across a diverse set of information needs and large populations. As a result, methods to understand searcher preferences solely via the retrospective analysis of logged search activity (e.g., which results or interface items receive the most attention given variations in ranking and/or interface presentation) are particularly attractive. More sophisticated experimental apparatuses allow different measures of systems performance to be computed and a more complete sense of searcher performance to be attained. There are also a number of alternatives to computing metrics from behavior, such as capturing labels directly from searchers directly (in situ) at search time, and measuring signals such as cognitive load and affect to enrich signals based only on the explicit actions that searchers perform.
8 - Interaction beyond the Individual
- from Part II - System SupportHelping People Search
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 249-266
-
- Chapter
- Export citation
-
Summary
People are important recipients and consumers of search outcomes, and can also be important participants in the search process – pursuing common search objectives and helping searchers answer questions. Many search systems model information seeking as a solitary process comprising interaction between human and machine. However, interactive searching occurs within a social context (Ingwersen and Järvelin, 2005). Frequently, the outcomes of searches are meant to inform interactions with others or to create resources that influence or are used by other people. In addition, searchers can benefit from other people's expertise and experiences. Conversations between individuals can assist with developing a plan and help searchers crystalize their thinking about a path forward.
As the world becomes more connected, interactions with other people are going to play an increasingly important role in search interaction. Search systems can play a central role in facilitating interpersonal interactions, both in terms of connecting people and in mediating their dialogs. They can connect strangers pursuing the same task at the same time (Bateman et al., 2012; González-Ibáñez et al., 2015), and facilitate co-searching between friends and colleagues alike (Morris and Horvitz, 2007; Amershi and Morris, 2008). The strangers may be search experts who can provide guidance on search strategies that would help people locate their information faster by better understanding how best to query, and/or they could be subject matter experts with the domain knowledge to answer searcher questions directly via synchronous or asynchronous communication channels without requiring further engagement with the search system (Adamic et al., 2008).
Social information can be employed by search systems in a number of ways, including: (1) asking others (e.g., social question and answer, where experts can help searchers answer questions); (2) working together (e.g., collaborative information seeking – including cases for which the collaboration may be initiated by the search system (González-Ibáñez et al. [2012]); (3) learning from the behavior of others in the aggregate (e.g., trails extracted from search logs can provide guidance on where searchers should go next; see Wexelblat and Maes [1999] and Joachims [2002]); (4) learning from link creation (e.g., hyperlink analysis of resources explicitly connected by document authors; see Brin and Page [1998] and Kleinberg [1999]), and; (5) explicit recommendations (e.g., recommender systems [see Konstan and Riedl, 2012]). Points (3), (4), and (5) are covered elsewhere in the book.
Part I - Background
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 13-14
-
- Chapter
- Export citation
-
Summary
In this part of the book, I provide background information about the collection and application of behavioral signals beyond what is already offered in the previous chapter. I discuss the ways in which evidence of search interaction can be collected and applied in search systems. Much of the previous research in this area has focused on the science behind information seeking behaviors or the effectiveness of novel interfaces to support those behaviors. In light of the data revolution, there has been a growth in interest in the collection, modeling, and application of search interaction behaviors at scale (across many thousands and millions of individuals) in settings such as web search and web browsing. In addition, there have been a number of recent technological advances in interaction methods (e.g. touch, gesture, and speech recognition), sensing, mobile and cloud computing, and machine learning that enable a range of new opportunities for search system design.
Dedication
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp v-vi
-
- Chapter
- Export citation
Preface
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp xi-xvi
-
- Chapter
- Export citation
-
Summary
Information seeking is a fundamental human activity, often conducted through interactions with automated search systems. The retrieval and comprehension of information returned by these systems is a key part of decision making and action in a broad range of settings; searching skills are now even taught in schools. The processes by which people retrieve and use information has been examined in detail by the information science, information retrieval, and human-computer interaction research communities for decades.
Information scientists have targeted the cognitive and behavioral mechanisms involved in the formulation of information needs and the processes by which people search for information and update their beliefs. The goal of searching is often regarded to be to reduce uncertainty in light of the information encountered, but there may also be the intention to increase that uncertainty, for example during exploratory or leisure search scenarios. Information retrieval researchers have targeted the development of new search technologies, including more advanced methods for ranking, indexing, and crawling, that facilitates the collection and selection of potentially relevant content from large document collections such as the World Wide Web or within large enterprises (where the goal may be to locate people with specific expertise rather than find information items). Human-computer interaction researchers have investigated how people interact with technology, and they have developed interfaces to allow searchers to explore and make sense of information resources as well as generate hypotheses to guide future exploration activities and decision making. In this book I discuss how new interaction capabilities such as touch and gesture, the emergence of cloud and mobile computing, machine learning, and big (and small) data mining will change the search landscape over the next decade and beyond. By building on these and other pillars, next-generation search systems will empower people and support the activities that they value.
This is the first book devoted to discussing how the range of emerging technologies can be employed to improve the search experience. To enable this transformation, many research communities – including information retrieval, human factors, data mining, and machine learning – must cooperate on the development of systems that empower searchers and leverage the broad array of tools at their disposal to make search a productive and pleasurable experience.
6 - Exploration, Complexity, and Discovery
- from Part II - System SupportHelping People Search
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 201-230
-
- Chapter
- Export citation
-
Summary
As the range of search tasks that people perform on search engines increases, supporting searchers in examining results, exploring result spaces, and discovering new insights from data will become key requirements in the design of search systems. An important part of providing this support is to identify when searchers are exploring and to design search interfaces to support exploratory activities.
In Chapter 5, I mainly focused on searching for information as an activity that search systems should support directly, and I discussed some of the existing and emerging techniques to support searchers in performing those searches. The focus for much of the work covered were cases for which the target of the search is known. Although search engines continue to improve how they handle lookup-based searches (assisted to a large extent by the availability of behavioral data), support for ill-defined, exploratory scenarios needs to be expanded. There is a range of information tasks for which information goals are less clearly defined; those tasks are built around searchers’ desires to explore (e.g., Marchionini 2006a; White and Roth, 2009). As discussed in detail in Chapter 4, exploration is an important aspect of information behavior that can be motivated by intrinsic and extrinsic motivations (Berlyne, 1960). The exploration of information spaces can yield key new insights and advances, promote serendipity, and foster creativity. Exploration is central in activities such as learning, understanding, and decision making. All of these outcomes may occur during the search for known items, but the attainment of such goals is not a primary objective in traditional search systems. However, searchers’ expectations are evolving and next-generation search system designers need to consider exploration and discovery as central elements of the search process.
In this chapter, I focus on both exploratory and complex search tasks, which have many of the following characteristics:
• Emphasis on learning and discovery, driven by task requirements and/or a desire to learn;
• Ill-defined and multi-aspect;
• Require both search and browsing;
• Higher-level goals beyond finding, such as enhanced understanding and decision-support; and
• Long-term (searches over multiple sessions) and persistent (even if searcher is not actively searching).
Exploration can range from general exploratory behavior (as defined in Section 4.1) to behavior associated with exploratory or complex search tasks, which is a more directed action (Marchionini, 1995).
3 - Modeling Interests and Intentions
- from Part I - Background
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 59-96
-
- Chapter
- Export citation
-
Summary
In this chapter, I discuss various aspects of the interactive search process, including next-generation search experiences and emerging trends Section 3.1). I discuss various models of search interaction that have been developed for applications such as click prediction, satisfaction, and relevance (Section 3.2). I also enumerate some of the main components required for model building, including data, data mining, and machine learning (Section 3.3). From the breadth of the topics covered in this chapter, clear that many factors must be considered in modeling interests and intentions via interactions with search systems.
Modeling Next-Generation Search Interaction
Let us begin with a high-level model of next-generation search interaction that reflects emerging trends in the area, yet builds on much of the work on collecting and representing search interaction that was described in Chapter 2. The model is depicted visually in Figure 3.1. Although an interaction model is not strictly necessary for a discussion of progress in this area, it can be useful in framing many of the contributions that are mentioned in this book. I discuss emerging trends likely to affect search interaction, as well as other factors including the role of large-scale behavioral data in guiding effective decisions by future searchers and search providers; generic and personalized machine-learned models of searchers’ interests, intentions, and search satisfaction levels; support for task completion; cloud-based application and storage (used to retain found information items that were found, generated by the searcher, or in the process of being generated (work in progress), as well as rapidly accessible profiles of searchers’ long-term interests and intentions – reflecting completed and ongoing search tasks); and context of various forms, natural interaction with search systems, and ubiquitous search (through mobile computing and support for cross-device interactions). Core elements of next-generation search interactions are proactive and reactive experiences and intelligent personal assistants working in concert to surface relevant and useful information at an appropriate time.
Emerging Trends
There are a number of emerging trends in computer and information sciences, and in society more broadly, that search engine designers need to consider when designing next-generation search systems. These will influence the design of search technologies, how people interact with search systems, searcher experiences, and search system evaluation. Some of the main trends can be summarized as follows:
12 - Data, Tools, and Privacy
- from Part III - Evaluation
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 361-384
-
- Chapter
- Export citation
-
Summary
An important aspect of many of the methods covered in this book is the availability of data on how people interact with search systems. It is therefore important to discuss how searcher data are collected, and what data are available for research purposes. An important aspect in mining, analyzing, and applying these data is searcher privacy, which permeates all aspects of collection and use – from the consent of searchers to collect the data at the outset, to the de-identification, aggregation, and restrictions of sharing and applying data (Horvitz and Mulligan, 2015). The collection of such interaction data is standard practice for large commercial entities, such as Web search engines, who use the data to understand how people are interacting with their services and improve the user experience. Because of privacy concerns, once the data are collected, they are usually not shareable with external parties. Efforts to release data (e.g., by America Online in 2006) have led to serious privacy breaches associated with a failure to completely anonymize the dataset. Serious events such as this make future broad data releases unlikely. Limited releases under license to researchers and the extreme anonymization of datasets have been used as strategies to address privacy challenges and promote research into behavioral analysis and user modeling.
In this chapter, I discuss the need for the shared resources (e.g., datasets), tools (e.g., logging support), and infrastructure that are necessary to build and evaluate competitive search systems. These pillars are important when comparing or coordinating the performance of interactive search systems across multiple experimental sites. Lagergren and Over (1998) described an experimental design for cross-site comparisons of experimental results (i.e., a matrix design to which participating sites must strictly adhere) to address issues such as two-way interactions and effects specific to how the experiment was conducted at a particular site, in the context of the TREC Interactive Track (in which a single search system was used as a baseline at all sites). This involved significant coordination effort and was still focused on comparing systems. Important alternative goals include advancing our understanding of search behavior, improving the design of systems to support searching, and facilitating comparability between laboratory studies performed at different sites.
2 - Collecting and Representing Search Interaction
- from Part I - Background
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 15-58
-
- Chapter
- Export citation
-
Summary
In mainstream search systems, search interactions typically assume the form of search queries and result selections. This information can be useful as implicit feedback to improve search performance when aggregated across many searchers (Joachims, 2002; Agichtein et al., 2006). Implicit feedback can also be collected at a personal level and used to update the search experience directly in real time (e.g., rearranging available information [White et al., 2005a; White et al., 2005b]), or to tailor the results for the current query using personalization or contextualization (Dou et al., 2007; Teevan et al., 2011b; Bennett et al., 2012). Trails that people follow through document collections (Bush, 1945; Bilenko and White, 2008) and trail destinations (White et al., 2007) can also be employed to help searchers understand the corpus and the domain (White and Huang, 2010), and ultimately to attain their search goals. In addition to being used to refine search results or other features (such as related searches or query auto-completions [shown dynamically as searchers compose queries]), recorded traces of search interaction can also be used as a diagnostic tool to understand when searchers are satisfied or frustrated with their search experience (Hassan et al., 2010; Aula et al., 2010; Feild et al., 2010), and consequently, where the search system could improve its performance or provide missing search support. Sections 2.12.3 discuss components of the search interaction process, including SERP interactions and sequences of actions extending beyond the search engine and into the corpus being searched.
There is a growing set of opportunities for search engines to learn from aggregated search activity as a new range of interface features emerge to integrate new capabilities, such as touch and gesture. Section 2.4 discusses mechanisms for collecting and representing interests and intentions beyond queries and clicks. Other advances, such as eye-gaze tracking and spoken dialog, will change both the manner and the settings in which people interact with search systems. Because interactions depend on both software and hardware, device-dependent models of relevance may be needed to accurately associate interaction events on each device with searchers’ common interests and intentions. Some progress has been made on modeling relevance on mobile devices using touch interactions (Guo et al., 2013), but more research in this area is anticipated.
7 - Learning and Use
- from Part II - System SupportHelping People Search
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 231-248
-
- Chapter
- Export citation
-
Summary
Learning involves the acquisition of new, or the modification of existing, knowledge, skills, and behaviors. This chapter focuses on the use of search systems to support learning-related activities. I discuss aspects of the psychology of learning, and how people apply what they have learned during the information-seeking process. Modeling and supporting learning is essential in next-generation search technology. Support for learning will become an important aspect of how we design search systems, as well as how we evaluate these systems to understand their utility. Search systems can help people learn new content and also be reminded about content accessed historically. Providing the context of previous learning episodes and helping restore searcher state and refresh their understanding of the current search topic, if it is not in a subject area reflective of their regular interests.
Searching is similar in many ways to learning (Schmeck, 1988; Davis and Palladino, 1995). Previous work has provided some basis for a strong connection between searching and learning based on the construction of enhanced knowledge structures as a person assimilates new information attained during the search process with existing knowledge (Wittrock, 1974; Yankelovich et al., 1985). Case (2002) states that sense-making (such as that described in Chapter 4) is theoretically grounded in the constructivist learning theories of Dewey (1933), who in turn argued that learning can only occur through the activity of problem solving. The psychology community has explored learning in detail (e.g., Piaget [1952] and Vygotsky [1962] studied learning in the context of childhood development). Systems such as SuperBook (Egan et al., 1989) and SuperManual (Folz and Landauer, 2007) have improved the usability of existing content, books and manuals respectively, via computer-based enhancements such as rich indexing (to help address the vocabulary mismatch problem) and fisheye visualizations (Furnas, 1986) (to help people navigate and orient within text). Such enhancements can help people better comprehend text and generate better quality information artifacts in composition tasks. In creating systems where the focus is on learning, lessons can be drawn from the e-learning and intelligent tutoring communities (Corbett et al., 1997), such that searchers can be purposely engaged in sustained reasoning activities during browsing. Related work in the hypertext community, including research on the creation of guided tours (Trigg, 1988) through document collections, may also help people in support of learning and understanding.
Part II - System SupportHelping People Search
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 139-140
-
- Chapter
- Export citation
-
Summary
This part of the book describes how the methods, principles, and models described in the previous chapter can be applied to build systems capable of supporting people's search-related interactions. I discuss both the current state-of-the-art in supporting searching, and anticipated advances in search systems in light of newly emerging technologies. Examples of existing search systems are included as appropriate.
4 - Models and Frameworks for Information Seeking
- from Part I - Background
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 97-138
-
- Chapter
- Export citation
-
Summary
The theory and practice of information seeking has been researched extensively in the Information Science and Information Retrieval (IR) research communities. Many of the models of interests and intentions described in the previous chapter have foundations in this literature. As such, it is worth reviewing the research in this area and how it applies to next-generation search systems. Many models of search behavior that have been proposed are based on observations of how people search on their own and how they interact with intermediaries, such as reference librarians, during the search process. Indeed, the reference librarian model (of a human search expert trying to satisfy a patron's information needs) remains the prevalent interaction model in many search systems, including commercial Web search engines. The primary difference is that in these systems human librarians have been largely replaced by automation in the retrieval process (including formulating effective queries via tools such as query auto-completion, query suggestion, and backend query alterations), and by the searcher themselves (for example, in decisions regarding the relevance, filtering, and synthesis of the retrieved items) to generate a set of relevant information items, and ultimately, one or more answers to the questions that motivated their search.
Information-seeking behavior forms part of the broader field of Information Behavior, which includes both intentional information seeking (such as querying or actively browsing for information), and unintentional behaviors (such as passively watching a television commercial; see Wilson [1999]). Models of information seeking are informative for the design of search interfaces and behavioral models (for tasks such as click prediction, satisfaction modeling, etc.) that meet searchers’ needs and are representative of the types of actions that they perform. One such conceptualization involves different modes of information searching, reflecting different ways that people search for information. Table 4.1 summarizes these different modes as based on Wilson (1998) and Bates (2005), which varies how directed (goal-oriented) and how active the searcher will be during the information search process.
Research in areas such as exploratory search, information foraging, sensemaking, learning, creativity, and knowledge discovery, alongside more traditional models of information seeking (e.g., cognitive, strategic, process, episodic, and stratified models of search interaction), have covered many of the modes highlighted in Table 4.1. There are a number of theories and frameworks that have contrasted searching and browsing (Belkin et al., 1993; Marchionini, 1995).
Notes
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 411-414
-
- Chapter
- Export citation
9 - Personalization and Contextualization
- from Part II - System SupportHelping People Search
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 267-304
-
- Chapter
- Export citation
-
Summary
Earlier in the book, I discussed modeling various aspects of search behavior, such as the relevance of search results or searcher satisfaction with search systems. In this chapter, the focus turns to the process of modeling searchers’ long-term interests (via personalization) and their current search situation (via contextualization). The personalization and contextualization of the user experience are increasingly important in search and recommendation systems, and enriching queries with this information facilitates the provision of individualized and situation-dependent experiences that will be commonplace in next-generation search systems, and are increasingly evident in today's search systems.
Information about searcher's interests and intentions can be used to tailor the search experience to individual searchers and to those in similar situations. Although limited to what situational information is visible to the search engine and affected by other limitations (such as profile size and log volume), rich models of search interests and their search situations can still be developed. In this chapter, I cover both personalization and contextualization; these concepts are often conflated in the research literature, they are but they are in fact quite different. I distinguish between them primarily in terms of the nature of the data used for model construction. Specifically, I define personalization and contextualization as follows:
• Personalization: Tailored to the individual searcher. Search systems may not be representative of individual searcher's information needs (Teevan et al., 2010), and personalization can help address this issue. Methods to support personalization are usually developed by modeling the long-term activity of individual searchers (e.g., search queries and clicks over a period of thirty days or more) to truly understand their interests. It is only by monitoring search behavior longitudinally that a truly individualized user profile can be constructed. Although some authors have regarded using within-session behavior as personalization (Daoud et al., 2009; Sriram et al., 2004), the data are often too sparse (just a few queries and document selections), and too task-specific, to adequately represent the searcher so that the search experience can be tailored to them. Because they rely on short-term interaction histories, such session-based models cannot impact the first query in the session (Bennett et el., 2012). […]
10 - Evaluation Measures
- from Part III - Evaluation
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 307-336
-
- Chapter
- Export citation
-
Summary
Measuring the performance of search systems is essential in improving their effectiveness. Computing measures (or metrics, used synonymously in this chapter) lets search providers benchmark current performance, as well as quantify the impact of any changes. Some measures target the outcomes of the search process (e.g., the relevance of the found items), while some are more focused on the search process itself (e.g., the efficiency or cognitive load of the search process). Although numerous measures of search system performance have been proposed, none can fully evaluate search systems from all perspectives. As search systems become more sophisticated and support a broader range of tasks, new evaluation metrics and metric combinations will be needed.
Engelbart (1962, p. 1) suggested that the increased capability attributable to augmenting human intellect would likely lead to: “more-rapid comprehension, better comprehension, the possibility of gaining a useful degree of comprehension in a situation that previously was too complex, speedier solutions, better solutions, and the possibility of finding solutions to problems that could not previously be solved.” Systems that offer such opportunities cannot simply be evaluated using traditional retrieval measures such as precision and recall, which only consider the relevance of the found content and how much of the relevant content is found. In this case, we would need metrics that assess the quality of the solution and assess the impact of the search process on people's understanding of the subject matter.
There are two groups of metrics considered in this chapter: (1) those that assess the search process in which the searcher was engaged; and (2) those that target the outcomes attained as a result of that process. For completeness, I cover some of the traditional metrics, but many of those discussed draw on research in other communities, such as psychology. Irrespective of the target for the metric computation, with enhancements in next-generation search systems, evaluation metrics (and methods, discussed in the Chapter 11) need to cater to a diverse range of searchers, tasks, and interactivity.
Traditionally, the unit of retrieval evaluation is the search query. Next-generation search systems, however, place an emphasis on supporting the completion of complete search tasks end-to-end (rather than considering queries independently and satisfying task-relevant information needs one query at a time).
Index
- Ryen W. White
-
- Book:
- Interactions with Search Systems
- Published online:
- 05 March 2016
- Print publication:
- 14 March 2016, pp 499-510
-
- Chapter
- Export citation