Dissertation Committee Members Research Articles

“It answers questions that I didn’t know I had”: PhD students’ evaluation of an information-sharing knowledge graph

PurposeInterdisciplinary PhD programs can be challenging as the vital information needed by students may not be readily available; it is scattered across the university’s websites, while tacit knowledge can be obtained only by interacting with people. Hence, there is a need to develop a knowledge management model to create, query and maintain a knowledge repository for interdisciplinary students. This study aims to propose a knowledge graph containing information on critical categories (faculty, classes, dissertations, etc.) and their relationships, extracted from multiple sources, essential for interdisciplinary PhD students. This study evaluates the usability of a participatory-designed knowledge graph intended to facilitate information exchange and decision-making.MethodologyThe authors used data from multiple sources (such as university websites, faculty profiles, publication and dissertation metadata and crowdsourced data) to generate a knowledge graph in the Neo4J Bloom platform. The authors recruited 15 interdisciplinary PhD students using convenience sampling from the University of Hawaiʻi at Mānoa at various PhD stages to design and populate the knowledge graph. Next, the authors conducted a mixed methods study to perform its usability evaluation. First, the authors engaged the students in a participatory design workshop to identify relevant graph queries. Second, the authors conducted semi-structured interviews to determine the usability of the knowledge graph and rate the queries. Each interview was coded with structural and thematic codes and was further analyzed using sentiment analysis in R programming language.FindingsThe usability findings demonstrate that interaction with this knowledge graph benefits PhD students by notably reducing uncertainty and academic stress, particularly among newcomers. Knowledge graph supported them in decision-making, especially when choosing collaborators (e.g. supervisor or dissertation committee members) in an interdisciplinary setting. Key helpful features are related to exploring student–faculty networks, milestones tracking, rapid access to aggregated data and insights into crowdsourced fellow students’ activities. However, they showed concerns about crowdsourced data privacy and accessibility. Although participants expressed the need for more qualitative data in the graph, they noted it helped identify people to talk to about the topics of their interest.OriginalityThe knowledge graph provides a solution to meet the personalized needs of doctoral researchers and has the potential to improve the information discovery and decision-making process substantially. It also includes the tacit knowledge exchange support missing from most current approaches, which is critical for this population and establishing interdisciplinary collaborations. This approach can be applied to other interdisciplinary programs and domains globally.

Methodological Quality and Publication of Doctoral Dissertations in Education: An Evaluative Study of Ten Years of K-12 Doctoral Dissertations in the United States

Aim/Purpose: Despite an increase in the numbers of K-12 educators pursuing doctoral degrees, it is unclear if the field of education has been significantly impacted by the research resulting from their doctoral dissertations. Accordingly, the quality of doctoral programs and dissertations and rate of publication after defense, warrants examination. Background: There have long been discussions regarding concerns about the quality of doctoral dissertations that come in concert with an increase in degrees awarded. This paper presents findings from a study examining dissertation quality for K-12 education doctorates awarded over a ten-year period in the United States. Methodology: The researchers randomly selected and evaluated the methodological quality of more than 500 dissertations during the last ten years from the Proquest dissertation database using a rubric adapted from Ronau et al. (2014). Using this rubric, the team described the quality of dissertations with respect to sampling, design, and validity. Contribution: A new rubric was developed and tested to assess the quality of quantitative, qualitative, and mixed methods studies. These results provided large-scale data on dissertation quality that may be used to guide doctoral programs in implementing program changes. Findings: Although the majority of the dissertations studied used qualitative methods of inquiry, quantitative research had significantly higher quality scores. Regardless of the methods used, many studies failed to employ rigorous validity controls. Recommendations for Practitioners: Doctoral programs may wish to make use of these results to improve programming, particularly in methods courses. These results may also be used to guide dissertation advisors and dissertation committee members, along with increasing publication rates among doctoral students. Recommendation for Researchers: Researchers may wish to extend this study by looking more closely at the methodological quality in specific subject areas. Research is also needed to improve the rubric developed for this study further and to determine other metrics of impact beyond publication. Impact on Society: Stronger methods sections may lead to more publications of dissertation studies, which can improve impact on educational practice and evidence-based decision-making. Future Research: Future research should assess the quality of research in the context of usefulness, impact on practice beyond publication, and topic currency. Further, the motives behind publishing dissertation work in fields that do not require publication should be examined.

The effectiveness of the PIUSE platform in research management and its relationship with the intensive use of the internet by thesis committee members and advisors

The effectiveness of the PIUSE platform in research management and its relationship with the intensive use of the internet by thesis committee members and advisors

Tom and Thode’s excellent adventure: Practical guidance on surviving seedy bars, preparing for a DJ career, dismantling the U.S. Panama Canal zone, and other useful tips when working with towed acoustic arrays in the Eastern Tropical Pacific

Tom Norris used many acoustic tools in his research career, but it is safe to say that towed passive acoustic arrays were one of his specialties. In 1998 the National Marine Fisheries Service conducted the Stenella Population and Abundance Monitoring (SPAM) cruise. The NOAA ship Endeavor was to survey enormous dolphin populations in the Eastern Tropical Pacific (ETP), starting from Lima, Peru, and ending in Panama City, Panama. Jay Barlow, one of my PhD thesis committee members, permitted me to participate in the cruise as a member of his towed array passive acoustic monitoring team. The first time I ever met Tom, the acoustics team leader, was when he was carrying a surfboard into his cabin, and I wondered how in the world he had managed to ship a surfboard into Peru. It was the first of many times I have enjoyed both working with and wondering about Tom, and our first trip together taught me practical things they forgot to teach in graduate school. Here, I will review Tom’s tips on Mackie mixer boards, buying drinks in dangerous bars, tracking dolphin whistles, not antagonizing visual observers, and liberating furniture from the U.S. Panama canal zone.

Enhancing big data in the social sciences with crowdsourcing: Data augmentation practices, techniques, and opportunities.

Proponents of big data claim it will fuel a social research revolution, but skeptics challenge its reliability and decontextualization. The largest subset of big data is not designed for social research. Data augmentation–systematic assessment of measurement against known quantities and expansion of extant data with new information–is an important tool to maximize such data's validity and research value. Using trained research assistants or specialized algorithms are common approaches to augmentation but may not scale to big data or appease skeptics. We consider a third alternative: data augmentation with online crowdsourcing. Three empirical cases illustrate strengths and limitations of crowdsourcing, using Amazon Mechanical Turk to verify automated coding, link online databases, and gather data on online resources. Using these, we develop best practice guidelines and a reporting template to enhance reproducibility. Carefully designed, correctly applied, and rigorously documented crowdsourcing help address concerns about big data's usefulness for social research.

In Memoriam: Edward Branigan

Teacher. Mentor. Dissertation committee member. Advocate. Colleague. Friend. These are the many roles that Ed Branigan filled in my life over the eleven-plus years I was privileged to know him. However, merely listing these roles does not really do justice to his impact on me, because it leaves out the kindness, generosity, wit, and enthusiasm that he always had in store for me in all of our interactions, be they post-lecture dinners together in Santa Barbara, movie marathons at his house in Oak Park, California, or, as was more and more common over the last few years, e-mail messages.

Seeing my science clearly.

“You are about to witness a thesis implode,” I declared to my summer student. As I ran the critical sample through the flow cytometer, my trembling hand betrayed my nervousness that the result would confirm my worst fears: I had been fooling myself. Three years prior, I had been a bright-eyed first-year Ph.D. student. “This is guaranteed to give you data,” my mentor had said of the project he wanted me to take over. Preliminary data suggested that a poorly understood protein might trigger immune cells to kill cancer cells. It was an exciting possibility, but I was concerned. The previous rotation student had been unable to replicate the finding and wasn't convinced the protein did anything at all. But I couldn't turn down a project that sounded so promising. ![][1] ILLUSTRATION: ROBERT NEUBECKER > “I had been spinning a story … instead of critically examining the facts.” To my relief, a new batch of protein I prepared yielded striking results. The science unfurled into a thrilling jigsaw puzzle. Beautiful flow cytometry plots revealed that our protein induced immune cells to mature. In tissue culture experiments, these cells killed cancer cells with startling ferocity. The debut in vivo study showed that mice treated with the protein had only a fraction of the black melanoma spots we found in controls. I presented the results—collected on Christmas Eve, no less—at our weekly lab meeting with a mother's pride: A causes B causes C, in perfect alignment with our neatly constructed hypotheses. The results were so astounding, in fact, that they began to make me suspicious. I—along with my thesis committee members and adviser—wondered whether my results could be due to contamination from the bacterial system I used to produce the protein. After all, it seemed logical that a fragment of a pathogen, not our protein, could be what triggered the immune cells. However, I was soon reassured. I tested for contamination, looking for global immune activation, and everything looked clean. Relieved, I carried on. But about a year later, a reviewer for a grant application identified a flaw in the control I had put my trust in. It was a largely unknown quirk in immunology, heretofore untested because my committee members—like me and my adviser—saw the logical sense of the story I had synthesized. That is what led to the fateful moment at the flow cytometer, where I was testing a commercial sample of my protein certified to be free of any bacterial contaminants—a crucial control that I had previously been too blinded by preconceived expectations to devise. I was aghast when I saw that my protein in this pure state did nothing. My rotation student glanced at me, concerned. I chuckled darkly. I realized that I had been spinning a story from my science, interpreting my results based on my conception of what should be happening instead of critically examining the facts in front of me. I stepped away from the lab bench for a week and confronted my mistakes with brutal honesty that ended up being cathartic. My adviser did the same. I leaned on generous loved ones as I worked through what felt like the five stages of grief. At first, I lamented the loss of my tidy, compelling story, but I eventually realized that element of my grief was misplaced. As scientists, we seek to understand nature's crooked truth. To simplify it—as tempting as that can be—is contrary to our quest to do meaningful science. It took me 6 months to recover and find closure by taking on a new project. My adviser supported my new direction, and eventually I graduated with my Ph.D. I was burned, but the scars remind me to be a better scientist. Nowadays, I don't stop at controls. I really ponder my results. Could the cause be something other than what I expect? I am slower to draw easy conclusions and instead make sure to consider every possible variable. I have learned to detach myself from what I want my science to be, and instead allow the science to lead me to understanding what actually is. I listen to my doubts, no matter how much the puzzle fits and “makes sense” at first—or 20th—glance. [1]: /embed/graphic-1.gif

The Honors Project Takes Shape: One Student’s Reflections on Being Gifted

In this article, an honors college student reflects on being gifted. In reflecting on her giftedness, she expresses her conceptions of various characteristics of giftedness that she has experienced through drawing. These drawings and associated captions are featured in this text, and the student discusses her process of experiencing the characteristics and developing the artworks. One of her honors thesis committee members provides connections to the literature for each of the issues the student discusses.

An Examination of the Fundamental Reasons Why Frank Ramsey Failed in His Reviews of J M Keynes's a Treatise on Probability

Frank P Ramsey did not consider the possibility of representing the concept of probability by an interval valued approach in his lifetime. Ramsey considered probability to be either ordinal or numerical. There was absolutely no room for interval estimates and interval probability in his approach. It is mentioned nowhere in any of his published or unpublished work. Ramsey was not familiar with the applied work done by George Boole on interval valued probability. Ramsey believed only in the standard mathematical theory of probability. The crucial assumption for Ramsey was additivity. Of course, Keynes rejected additivity except as a special case. This is the major difference between Ramsey and Keynes. Keynes believed it obvious that there is missing relevant evidence or knowledge in many decisions made by human beings. Non addiitvity immediately leads to uncertainty. Uncertainty requires the existence of non additivity. Keynes made it explicit in Part II of the A Treatise on Probability that additivity was a special case and not the general case. Unfortunately, Ramsey could not follow Keynes’s analysis in Part II of the TP. This, for instance, was a severe problem also for both F Y Edgeworth and E B Wilson. However, Edgeworth figured out that Keynes’s “non numerical” probabilities were interval valued by means of a very careful reading of Chapter III of the TP. It is very probable that Edgeworth also had the benefit of many private discussions with Keynes between 1909 and 1920 when both men were co-editors of the Economic Journal. Unfortunately, Ramsey’s conclusion was that Keynes’s “non numerical probabilities” were ordinal probabilities. Economists have mistakenly followed Ramsey’s assessment and ignored Edgeworth’s vastly superior assessment of the interval valued nature of Keynesian probability. The major defect in the CWJMK version of the A Treatise on Probability is that the editorial foreword was written by a rabid proponent of Ramsey’s belief in additivity, Richard Braithwaite. Anyone reading this editorial foreword before he/she reads the A Treatise on Probability will be completely biased against Keynes’s technical analysis in the A Treatise on Probability. The only existing antidote to Braithwaite’s “review” is the two combined reviews of Edgeworth, who provided, overall, the best reviews ever written of the A Treatise on Probability and Bertrand Russell’s review. In fact, I had to use Edgeworth’s two reviews to buttress my case in support of my dissertation with one of my dissertation committee members.

Dissertation Proposal Defense - "High-Tech Style:" A Student's Perspective on Another Use of Distance Learning Technology

Most discussions of the increased use of distance learning focus on using distance learning technology for providing academic courses, a natural response to the high number of students taking distance learning courses. In fact, by this year (2000) it is expected that 11.6 million students will have taken one or more distance learning courses (Goldberg, 1998 in Banas and Emory, 1998). However, there are other uses for this technology within the educational setting. One such use of distance learning technology in the graducate educational setting is for dissertation proposal defenses, particularly when dissertation committee members are located on university campuses that are geographically separated. This paper is a case study describing such an event: a dissertation proposal defense facilitated by the use of interactive television (I-TV) distance learning technology. Included in this case study is the presentation of a student's perspective on using distance learning technology, supplemented by faculty responses to a survey distributed by the student soliciting feedback on their experiences in participating in the I-TV facilitated proposal defense.

Enhancing Big Data in the Social Sciences with Crowdsourcing: Data Augmentation Practices, Techniques, and Opportunities

Some claim that “Big Data” will fuel a revolution in the social sciences, while skeptics challenge Big Data as unreliably measured, decontextualized, and lacking content validity. We argue that Big Data projects can be enhanced through data augmentation with crowdsourcing marketplaces like Amazon Mechanical Turk (MTurk). Following a content analysis of academic applications of MTurk, we present three empirical cases to illustrate the strengths and limits of crowdsourcing and address social science skepticism. The case studies use MTurk to (1) verify machine coding of the academic discipline of dissertation committee members, (2) link online product pages to an online book database, and (3) gather data on mental health resources at colleges. We consider the costs and benefits of augmenting Big Data with crowdsourcing marketplaces and provide guidelines on best practices. We also offer a standardized reporting template that will enhance reproducibility. This study expands the use of micro-task marketplaces to enhance social science acceptance of Big Data.

Event-based Multi-document Summarization

Daily amount of news reporting real-world events is growing exponentially. At the same time, Organizations are looking for information about current and past events that affects them, such as mergers and acquisitions of companies. The Organizations need to obtain event information in a fast and summarized form to make decisions. Event-based retrieval and summarization systems offer an efficient solution to this problem. Most summarization research work uses news stories. Although this type of documents is characterized by conveying information about events, almost all work concentrates on approaches that do not take into account this aspect. The proposed multi-document summarization methods are based on the hierarchical combination of single-document summaries. We improved our multi-document summarization methods using event information. Our approach is based on a two-stage single-document\ method that extracts a collection of key phrases, which are then used in a centrality-asrelevance passage retrieval model. To adapt centrality-as-relevance single-document summarization for multi-document summarization that is able to use event information, we needed a good and adaptable baseline system. Because the key phrase extraction play a significant role in the summarization, we improved a state-of-the-art key phrase extraction toolkit using four additional sets of semantic features. The event detection method is based on Fuzzy Fingerprint, which is a supervised method trained on documents with annotated event tags. We explored three different ways to integrate event information, achieving state-of-the-art results in both single and multi-document summarization using filtering and event-based features. To cope with the possible usage of different terms to describe the same event, we explored distributed representations of text in the form of word embeddings, which contributed to improve the multi-document summarization results. The automatic evaluation and user study performed show that these methods improve upon current state-of-the-art multi-document summarization systems on two mainstream evaluation datasets, DUC 2007 and TAC 2009. We show a relative improvement in ROUGE- 1 scores of 16% for TAC 2009 and of 17% for DUC 2007. We have also obtained improvements in ROUGE-1 upon current state-of-the-art single-document summarization systems of between 32% in clean data and 19% in noisy data. These improvements derived from the inclusion of key phrases and event information. The extraction of key phrases was also refined with additional pre-processing steps and features, which lead to a relative improvement in NDCG scores of 9%. The introduction of Fuzzy Fingerprints for event detection enabled the detection of all event types, while the best competitor, an SVM with enhanced features, only detects roughly 85% of the different types of events. This lead to a large increase in the G-Mean and variants results when using the Fuzzy Fingerprints method. This doctoral work was carried under the Dual Ph.D. program in Languages and Information Technologies of Carnegie Mellon Portugal Program. The program took place at both the Language Technologies Institute at Carnegie Mellon University (CMU) and at the Department of Computer of Science and Engineering at Instituto Superior Técnico (IST). The doctoral work was performed under the supervision of University Professor Jaime Carbonell (CMU), Distinguished Career Professor Anatole Gershman (CMU), Assistant Professor David Martins de Matos (IST), and Assistant Professor Joo P. Neto (IST). Full Professor Ricardo Baeza-Yates (Yahoo!Research/Univ. Pompeu Fabra/Univ. de Chile), Research Professor Eduard Hovy (CMU), Associate Professor Ana Paiva (IST), and Full Professor Isabel Trancoso served as dissertation committee members.

Geo-Temporal Mining and Searching of Events from Web-based Image Collections

The proliferation of Web- and social media-based photo-sharing applications have not only opened many possibilities but also resulted in new needs and challenges. They have resulted in a large amount of personal photos being available for public access. One of the most interesting characteristics of these data is that they are surrounded by 1) textual annotations, also called tags, which are intended to describe and categorize, by collective user efforts, the uploaded resources 2) temporal information referring to when a pictures has been taken and often by 3) a locational information describing where the picture has been taken. Despite the recent developments and technological advances in Web-based mediasharing applications, the continuously increasing amount of available information has made the access to the photos a demanding task. In general, we can address this challenge by allowing photo collections to be organized and browsed through the concept of events. We also believe most users are familiar with searching photo collections using events as starting points. Aiming at supporting the detection and search of event-related photos, this thesis proposes a novel framework for extracting pictures related to real-life events from a collection of Web-images by leveraging on their temporal geographical and textual annotations and comparing the proposed approach with existing related state-of-the-art approaches. Second, a set of geographical features is proposed describing the characteristics of the geographical profile of query terms deriving concepts from exploratory analysis. Third, the thesis provides two different tag-based search framework to improve the effectiveness of searching images related to events. The first framework is based on temporal and geographical proximity of query terms to the temporal neighbours of a given timestamped query, while the second framework is based on a novel machine-learning based query expansion method combining the heterogeneous textual, temporal and geographical similarity between query terms and candidate expansion terms for the selection of the expansion terms given a free text textual query. All the proposed methods have been evaluated by performing extensive experiments on real data gathered from media-sharing applications on the Web. Where possible, comparison with related techniques has been performed to reinforce the validity of the presented approaches. The proposed methods have shown promising results in both the extraction and clustering of event-related images and searching event-related pictures by using metrics from the state of the art. This doctoral work has been performed at the Department of Computer and Information Science at the Norwegian University of Science and Technology (NTNU) under advise of Associate Professor Heri Ramampiaro and with co-supervisors Associate Professor Roger Midtstraum and Associate Professor Randi Karlsen. Professor Ramesh Jain (University of California Irvine), Assistant Professor Claudia Hauff (TU Delft) and Associate Professor Trond Aalberg (NTNU) served as dissertation committee members. Available online at: http://www.idi.ntnu.no/research/doctor_theses/ruocco.pdf.

Aportaciones de lectores académicos a la formación para la investigación educativa

Recibido 3 de setiembre de 2013 • Corregido 30 de enero de 2014 • Aceptado 2 de abril de 2014
 Este trabajo presenta una reflexión basada en elementos empíricos, derivados de la obra de Moreno, Torres y Jiménez (2014), titulada El discurso de los lectores académicos como mediación para formar investigadores en el posgrado, en la cual se reporta una investigación cuyo objeto de estudio es el contenido y orientación de los mensajes recibidos por estudiantes de doctorado en educación, de parte de formadores que realizaron lecturas críticas de sus avances de tesis doctoral. Se eligió este nivel educativo porque en México prácticamente todos los programas doctorales se encuentran orientados a la investigación. Por ende, se concibe que en el doctorado los estudiantes se forman para ser capaces de realizar investigación en un campo específico de conocimiento. Los datos, provenientes de 101 reportes escritos realizados por 9 lectores académicos a los trabajos de 60 estudiantes de dos programas de doctorado en educación en México, permiten dar cuenta de aspectos importantes que pueden aprender los estudiantes respecto a la forma de realizar investigación educativa y de reportarla en un texto escrito. En este trabajo tales aspectos se exponen en tres ejes de contenido, recuperando partes del discurso de los lectores.

‘A Little Respect’: An Inquiry-Driven Classroom Honors a Student's Right to Question

The practice of teaching can be characterized as a calling, a duty, and a gift. It can also be an opportunity for oppression — or for liberation from habits of oppression. In his role as dissertation chair and dissertation committee member, Dr Bruce encourages doctoral candidates to read about, practice, and reflect on new approaches to the relationship between teacher and student. As a teacher, a mentor, and a friend, he embodies the values he invites others to adopt. This article describes how one doctoral student's philosophy and practice of teaching changed because of the gift that Dr Bruce was and is still.

Improving the effectiveness of language modeling approaches to information retrieval

Improving the effectiveness of language modeling approaches to information retrieval

Plagiarism Among Dissertations: Prevalence at Online Institutions

The current research literature has claimed that plagiarism is a significant problem in postsecondary education. Unfortunately, these claims are primarily supported by self-report data from students. In fact little research has been done to quantify the prevalence of plagiarism particularly at the advanced graduate education level. Further, few studies exist on online education even though this is a rapidly growing sector of higher education. This descriptive study quantified the amount of plagiarism that existed among 100 doctoral dissertations that were published by institutions granting doctorate degrees through a primarily online format. The dissertations were submitted to Turnitin plagiarism detection software for analysis. The mean similarity index of these dissertations was 15.1 (SD = 13.02). The results were then categorized per previous research. Forty-six percent of the dissertations were classified as having a low level of plagiarism while 11 % had a medium level and 3 % had a high level. Further analysis revealed that 72 % of the dissertations had at least one case of improper paraphrasing and citation (verbatim text accompanied by a citation) and 46 % had verbatim text without any citation. The results of this study should encourage faculty, dissertation committee members, university administrators, and accrediting bodies to take action to help reduce the level of plagiarism among doctoral learners. Suggestions for future research include comparing online and brick-and-mortar dissertation plagiarism rates, a larger study to investigate plagiarism trend data, and surveys of faculty about how they address plagiarism and ethics during the dissertation process.

Document ranking with quantum probabilities

In this thesis we investigate the use of quantum probability theory for ranking documents. Quantum probability theory is used to estimate the probability of relevance of a document given a user's query. We posit that quantum probability theory can lead to a better estimation of the probability of a document being relevant to a user's query than the common IR approach, i. e. the Probability Ranking Principle (PRP), which is based upon Kolmogorovian probability theory. Following our hypothesis, we formulate an analogy between the document retrieval scenario and a physical scenario, that of the double slit experiment. Through the analogy, we propose a novel ranking approach, the quantum probability ranking principle (qPRP). Key to our proposal is the presence of quantum interference. Mathematically, this is the statistical deviation between empirical observations and expected values predicted by the Kolmogorovian rule of additivity of probabilities of disjoint events in configurations such that of the double slit experiment. While PRP explicitly assumes that the relevancy of a document is independent of that of other documents, we suggest that qPRP implicitly models interdependent document relevance through quantum interference and thus is suited to those document ranking tasks where the independence assumption fails. Throughout the thesis, we also suggest how quantum interference can be estimated for effective document ranking. To validate our proposal and to gain more insights about approaches for document ranking, we (1) analyse PRP, qPRP and other ranking approaches, exposing the assumptions underlying their ranking criteria and formulating the conditions for the optimality of the two ranking principles, (2) empirically compare three ranking principles (i. e. PRP, interactive PRP, and qPRP) and two state-of-the-art ranking strategies in two retrieval scenarios, those of ad-hoc retrieval and diversity retrieval, (3) analytically contrast the ranking criteria of the examined approaches, exposing similarities and differences, (4) study the ranking behaviours of approaches alternative to PRP in terms of the kinematics they impose on relevant documents, i. e. by considering the extent and direction of the movements of relevant documents across the ranking recorded when comparing PRP against its alternatives. Our findings show that the effectiveness of the examined ranking approaches strongly depends upon the evaluation context. In the traditional evaluation context of ad-hoc retrieval, PRP is empirically shown to be better than or comparable to alternative ranking approaches. However, when evaluation contexts that account for interdependent document relevance are examined (i. e. when the relevance of a document is assessed also with respect to other retrieved documents, as it is the case in the diversity retrieval scenario), the use of quantum probability theory and thus of qPRP is shown to improve retrieval and ranking effectiveness over the traditional PRP and alternative ranking strategies, such as Maximal Marginal Relevance, Portfolio theory, and Interactive PRP. This work represents a significant step forward regarding the use of quantum theory in information retrieval. It demonstrates that the application of quantum theory to problems within information retrieval can lead to improvements both in modelling power and retrieval effectiveness, allowing the constructions of models that capture the complexity of information retrieval situations. Furthermore, the thesis opens up a number of lines of future research. These include investigating estimations and approximations of quantum interference in qPRP, exploiting complex numbers for the representation of documents and queries, and applying the concepts underlying qPRP to tasks other than document ranking. This dissertation was completed at School of Computing Science, University of Glasgow under the advise of Dr. Leif Azzopardi and Prof. Keith van Rijsbergen. Prof. Norbert Fuhr, Dr. Iadh Ounis, and Dr. John O'Donnell served as dissertation committee members. For the full dissertation, visit: http://theses.gla.ac.uk/3463.

Efficient query processing in distributed search engines

Web search engines have to deal with a rapidly increasing amount of information, high query loads and tight performance constraints. The success of a search engine depends on the speed with which it answers queries (efficiency) and the quality of its answers (effectiveness). These two metrics have a large impact on the operational costs of the search engine and the overall user satisfaction, which determine the revenue of the search engine. In this context, any improvement in query processing efficiency can reduce the operational costs and improve user satisfaction, hence improve the overall benefit. In this thesis, we elaborate on query processing efficiency, address several problems within partitioned query processing, pruning and caching and propose several novel techniques: First, we look at term-wise partitioned indexes and address the main limitations of the state-of-the-art query processing methods. Our first approach combines the advantage of pipelined and traditional (non-pipelined) query processing. This approach assumes one disk access per posting list and traditional term-at-a-time processing. For the second approach, we follow an alternative direction and look at document-at-a-time processing of sub-queries and skipping. Subsequently, we present several skipping extensions to pipelined query processing, which as we show can improve the query processing performance and/or the quality of results. Then, we extend one of these methods with intra-query parallelism, which as we show can improve the performance at low query loads. Second, we look at skipping and pruning optimizations designed for a monolithic index. We present an efficient self-skipping inverted index designed for modern index compression methods and several query processing optimizations. We show that these optimizations can provide a significant speed-up compared to a full (non-pruned) evaluation and reduce the performance gap between disjunctive (OR) and conjunctive (AND) queries. We also propose a linear programming optimization that can further improve the I/O, decompression and computation efficiency of Max-Score. Third, we elaborate on caching in Web search engines in two independent contributions. First, we present an analytical model that finds the optimal split in a static memory-based two-level cache. Second, we present several strategies for selecting, ordering and scheduling prefetch queries and demonstrate that these can improve the efficiency and effectiveness of Web search engines. We carefully evaluate our ideas either using a real implementation or by simulation using real-world text collections and query logs. Most of the proposed techniques are found to improve the state-of-the-art in the conducted empirical studies. However, the implications and applicability of these techniques in practice need further evaluation in real-life settings. This dissertation was completed at the Department of Computer and Information Science at the Norwegian University of Science and Technology (NTNU) under advise of Prof. Svein Erik Bratsberg, Dr. Øystein Torbjørnsen and Dr. Magnus Lie Hetland. Some of the work was done in collaboration with Yahoo! Research Barcelona and mentored by Prof. Ricardo Baeza-Yates and Dr. B. Barla Cambazoglu. Prof. Alistair Moffat (University of Melbourne), Dr. Christina Lioma (University of Copenhagen) and Prof. Kjell Bratsbergsengen (NTNU) served as dissertation committee member. Available online at: http://www.idi.ntnu.no/research/doctor_theses/simonj.pdf.

Federated search in heterogeneous environments

In information retrieval, federated search is the problem of automatically searching across multiple distributed collections or resources. It is typically decomposed into two subsequent steps: deciding which resources to search ( resource selection ) and deciding how to combine results from multiple resources into a single presentation ( results merging ). Federated search occurs in different environments. This dissertation focuses on an environment that has not been deeply investigated in prior work. The growing heterogeneity of digital media and the broad range of user information needs that occur in today's world have given rise to a multitude of systems that specialize on a specific type of search task. Examples include search for news, images, video, local businesses, items for sale, and even social-media interactions. In the Web search domain, these specialized systems are called verticals and one important task for the Web search engine is the prediction and integration of relevant vertical content into the Web search results. This is known as aggregated web search and is the main focus on this dissertation. Providing a single-point of access to all these diverse systems requires federated search solutions that can support result-type and retrieval-algorithm heterogeneity . This type of heterogeneity violates major assumptions made by state-of-the-art resource selection and results merging methods and motivates the development of new techniques. While existing resource selection methods derive evidence exclusively from sampled resource content, the approaches proposed in this dissertation draw on machine learning as a means to easily integrate various different types of evidence. These include, for example, evidence derived from (sampled) vertical content, vertical query-traffic, click-through information, and properties of the query string. In order to operate in a heterogeneous environment, we focus on methods that can learn a vertical-specific relationship between features and relevance. We also present methods that reduce the need for human-produced training data. In particular, we focus on the situation where we have vertical-relevance judgments for some verticals and want to learn a predictive model for a vertical associated with no training data. Existing results merging methods formulate the task as score normalization. In a more heterogeneous environment, however, combining results into a single presentation requires satisfying a number of layout constraints. The dissertation proposes a novel formulation of the task: block ranking . During block-ranking, the objective is to rank sequences of results that must appear grouped together (vertically or horizontally) in the final presentation. Based on this formulation, the dissertation proposes and empirically validates a cost-effective methodology for evaluating aggregated web search results. Finally, it proposes the use of machine learning methods for the task of block-ranking. This dissertation was completed at School of Computer Science at Carnegie Mellon University under the advise of Dr. Jamie Callan (dissertation committee chair). Dr. Jaime Carbonell, Dr. Yiming Yang, and Dr. Fernando Diaz served as dissertation committee members. For the full dissertation, visit: http://ils.unc.edu/~jarguell.