Open-domain question answering relies on efficient passage retrieval to select candidate contexts, where traditional sparse vector space models, such as TF-IDF or BM25, are the de facto method. In this work, we show that retrieval can be practically implemented using dense representations alone, where embeddings are learned from a small number of questions and passages by a simple dual-encoder framework. When evaluated on a wide range of open-domain QA datasets, our dense retriever outperforms a strong Lucene-BM25 system greatly by 9%-19% absolute in terms of top-20 passage retrieval accuracy, and helps our end-to-end QA system establish new state-of-the-art on multiple open-domain QA benchmarks.

/pdf/dense-passage-retrieval-for-open-domain-question-answering-qq5rzovuru.pdf

Dense Passage Retrieval for Open-Domain Question Answering

This short communication uses a simple experiment to show that fitting to a power law distribution by using graphical methods based on linear fit on the log-log scale is biased and inaccurate. It shows that using maximum likelihood estimation (MLE) is far more robust. Finally, it presents a new table for performing the Kolmogorov-Smirnov test for goodness-of-fit tailored to power-law distributions in which the power-law exponent is estimated using MLE. The techniques presented here will advance the application of complex network theory by allowing reliable estimation of power-law models from data and further allowing quantitative assessment of goodness-of-fit of proposed power-law models to empirical data.

Problems with fitting to the power-law distribution

This paper provides overview and instruction regarding the evaluation of interactive information retrieval systems with users The primary goal of this article is to catalog and compile material related to this topic into a single source This article (1) provides historical background on the development of user-centered approaches to the evaluation of interactive information retrieval systems; (2) describes the major components of interactive information retrieval system evaluation; (3) describes different experimental designs and sampling strategies; (4) presents core instruments and data collection techniques and measures; (5) explains basic data analysis techniques; and (4) reviews and discusses previous studies This article also discusses validity and reliability issues with respect to both measures and methods, presents background information on research ethics and discusses some ethical issues which are specific to studies of interactive information retrieval (IIR) Finally, this article concludes with a discussion of outstanding challenges and future research directions

https://www.nowpublishers.com/article/DownloadSummary/INR-012

Methods for Evaluating Interactive Information Retrieval Systems with Users

Privacy enhancing technologies like OpenSSL, OpenVPN or Tor establish an encrypted tunnel that enables users to hide content and addresses of requested websites from external observers This protection is endangered by local traffic analysis attacks that allow an external, passive attacker between the PET system and the user to uncover the identity of the requested sites. However, existing proposals for such attacks are not practicable yet.We present a novel method that applies common text mining techniques to the normalised frequency distribution of observable IP packet sizes. Our classifier correctly identifies up to 97% of requests on a sample of 775 sites and over 300,000 real-world traffic dumps recorded over a two-month period. It outperforms previously known methods like Jaccard's classifier and Naive Bayes that neglect packet frequencies altogether or rely on absolute frequency values, respectively. Our method is system-agnostic: it can be used against any PET without alteration. Closed-world results indicate that many popular single-hop and even multi-hop systems like Tor and JonDonym are vulnerable against this general fingerprinting attack. Furthermore, we discuss important real-world issues, namely false alarms and the influence of the browser cache on accuracy.

/pdf/website-fingerprinting-attacking-popular-privacy-enhancing-1sej9sovca.pdf

Website fingerprinting: attacking popular privacy enhancing technologies with the multinomial naïve-bayes classifier

Collaboration is often a critical aspect of scientific research, which is dominated by complex problems, rapidly changing technology, dynamic growth of knowledge, and highly specialized areas of expertise. An individual scientist can seldom provide all of the expertise and resources necessary to address complex research problems. This paper describes collaboration among a group of scientists, and considers how their experiences are socially shaped. The scientists were members of a newly formed distributed, multi-disciplinary academic research center that was organized into four multi-disciplinary research groups. Each group had 14 to 34 members, including faculty, postdoctoral fellows and students, at four geographically dispersed universities. To investigate challenges that emerge in establishing scientific collaboration, data were collected about members' previous and current collaborative experiences, perceptions regarding collaboration, and work practices during the center's first year of operation. The data for the study includes interviews with members of the center, observations of videoconferences and meetings, and a center-wide sociometric survey. Data analysis has led to the development of a framework that identifies forms of collaboration that emerged among scientists (e.g., complementary and integrative collaboration) and associated factors, which influenced collaboration including personal compatibility, work connections, incentives, and infrastructure. These results may inform the specification of social and organizational practices, which are needed to establish collaboration in distributed, multi-disciplinary research centers.

/pdf/an-emerging-view-of-scientific-collaboration-scientists-10d75bnoyo.pdf

An emerging view of scientific collaboration: scientists' perspectives on collaboration and factors that impact collaboration

Changes in the topography of the Web can be expressed in at least four ways: (1) more sites on more servers in more places, (2) more pages and objects added to existing sites and pages, (3) changes in traffic, and (4) modifications to existing text, graphic, and other Web objects. This article does not address the first three factors (more sites, more pages, more traffic) in the growth of the Web. It focuses instead on changes to an existing set of Web documents. The article documents changes to an aging set of Web pages, first identified and collected in December 1996 and followed weekly thereafter. Results are reported through February 2001. The article addresses two related phenomena: (1) the life cycle of Web objects, and (2) changes to Web objects. These data reaffirm that the half-life of a Web page is approximately 2years. There is variation among Web pages by top-level domain and by page type (navigation, content). Web page content appears to stabilize over time; aging pages change less often than once they did.

Web page change and persistence---a four-year longitudinal study

This paper considers the status of information science as science through an exploration ofone of the leading journals in the field – the Journal of the American Society for InformationScience (JASIS) from its initial publication as American Documentation (AD) in 1950 through theclosing issue of its Silver Anniversary year in December 1999. It is a bibliometric examination ofAD/JASIS articles. Based on our analysis of articles published in AD and JASIS from 1950 to1999, we find that there has been a slow but perhaps inevitable shift based first on the single nonfundedresearcher and author to a much wider research and publishing participation amongauthors, regions, corporate authors, and countries. This suggests not only cross-fertilization ofideas, but also more complex research questions. A small trend toward greater external fundingfurther reinforces this hypothesis. Information may no longer be "little" science, but it is also not"big" science.

Information science as "Little Science":The implications of a bibliometric analysis of theJournal of the American Society for Information Science

We recognize that documents on the World Wide Web are ephemeral and changing. We also recognize that Web documents can be categorized along a number of dimensions, including “publisher,” size, object mix, as well as purpose, meaning, and content. This study is first a preliminary exploration into Web page and Web site mortality rates. It then considers two types of change: Content and structural. Finally, the study is concerned with understanding those constancy and permanence phenomena for different Web document classes. It is suggested that, from the perspective of information maintenance and retrieval, the WWW does not represent revolutionary change. In fact, in some ways the Web is a less sophisticated form than traditional publication practices. Finally, this study explores the “short memory” and “mind changing” of the World Wide Web.

An analysis of Web page and Web site constancy and permanence

A longitudinal study of Web pages continued: a consideration of document persistence.

L'auteur examine differents codes ethiques promulgues par les associations professionnelles, relatifs aux pratiques des specialistes de l'information, en pointant les contradictions de ces systemes de valeurs entre eux : primaute des droits du client individuel contre primaute des droits de l'institution, du bien public ou de l'interet national ; role de l'organisme professionnel dans la reglementation du comportement des specialistes de l'information. Il cherche ainsi a mettre a jour le coeur des principes applicables aux professions de l'information et propose en conclusion un modele de code base sur ces principes.

Wallace Koehler

Papers

Web page change and persistence---a four-year longitudinal study

Information science as "Little Science":The implications of a bibliometric analysis of theJournal of the American Society for Information Science

An analysis of Web page and Web site constancy and permanence

A longitudinal study of Web pages continued: a consideration of document persistence.

A Search for Core Values: Towards a Model Code of Ethics for Information Professionals.