Post ranking in a blogosphere

Traditional approaches to rule-based information extraction (IE) have primarily been based on regular expression grammars. However, these grammar-based systems have difficulty scaling to large data sets and large numbers of rules. Inspired by traditional database research, we propose an algebraic approach to rule-based IE that addresses these scalability issues through query optimization. The operators of our algebra are motivated by our experience in building several rule-based extraction programs over diverse data sets. We present the operators of our algebra and propose several optimization strategies motivated by the text-specific characteristics of our operators. Finally we validate the potential benefits of our approach by extensive experiments over real-world blog data.

Social networks need to manage and control the drift of insane amount of information by filtering and ranking everything in order to ensure they are right there for users' viewing pleasure. However, the realization of social networks ranking is currently dictated by fairly straightforward optimization algorithms. Hence, there is a need for a newly enhanced and improved ranking algorithm to be formulated since users have been occasionally seeing what they should not. In this paper, we conduct a user survey to gain a deeper insight on user needs for ranking algorithm. The outcome of the user survey is useful for the development of accurate and efficient ranking algorithm.

Motivated by the need to assess the concrete security of a device after a side channel attack, there has been a flurry of recent work designing both key rank and key enumeration algorithms. Two main competitors for key ranking can be found in the literature: a convolution based algorithm put forward by Glowacz et al. (FSE 2015), and a path counting based algorithm proposed by Martin et al. (Asiacrypt 2015). Both key ranking algorithms can be extended to key enumeration algorithms (Poussier et al. (CHES 2016) and Martin et al. (Asiacrypt 2015)). The two approaches were proposed independently, and have so far been treated as uniquely different techniques, with different levels of accuracy. However, we show that both approaches (for ranking) are mathematically equivalent for a suitable choice of their respective discretisation parameter. This settles questions about which one returns more accurate rankings. We then turn our attention to their related enumeration algorithms and determine why and how these algorithms differ in their practical performance.

Twitter has gained a significant prevalence among the users across the numerous domains, in the majority of the countries, and among different age groups. It servers a real-time micro-blogging service for communication and opinion sharing. Twitter is sharing its data for research and study purposes by exposing open APIs that make it the most suitable source of data for social media analytics. Applying data mining and machine learning techniques on tweets is gaining more and more interest. The most prominent enigma in social media analytics is to automatically identify and rank influencers. This research is aimed to detect the user's topics of interest in social media and rank them based on specific topics, domains, etc. Few hybrid parameters are also distinguished in this research based on the post's content, post’s metadata, user’s profile, and user's network feature to capture different aspects of being influential and used in the ranking algorithm. Results concluded that the proposed approach is well effective in both the classification and ranking of individuals in a cluster.

Twitter (and similar microblogging services) has become a central nexus for discussion of the topics of the day. Twitter data contains rich content and structured information on users' topics of interest and behavior patterns. Correctly analyzing and modeling Twitter data enables the prediction of the user behavior and preference in a variety of practical applications, such as tweet recommendation and followee recommendation. Although a number of models have been developed on Twitter data in prior work, most of these only model the tweets from users, while neglecting their valuable retweet information in the data. Models would enhance their predictive power by incorporating users' retweet content as well as their retweet behavior. In this paper, we propose two novel Bayesian nonparametric models, URM and UCM, on retweet data. Both of them are able to integrate the analysis of tweet text and users' retweet behavior in the same probabilistic framework. Moreover, they both jointly model users' interest in tweet and retweet. As nonparametric models, URM and UCM can automatically determine the parameters of the models based on input data, avoiding arbitrary parameter settings. Extensive experiments on real-world Twitter data show that both URM and UCM are superior to all the baselines, while UCM further outperforms URM, confirming the appropriateness of our models in retweet modeling.

web.media.mit.edu/∼ankit/bokode/ A bokode is a 3 mm wide circle of encoded data that can be decoded by a digital camera located up to several meters away. This is not commercially available technology, but it has the potential to replace barcodes and kindred encoding systems. The development team for this new technology is based at MIT and there are many added features beyond reading textual data (Fig. 1). Bokodes exploit the bokeh effect of ordinary camera's lenses, which means that optical phase data in the coded spot is readable when the camera is made to be deliberately out of focus (this is explained at the site). The value-added aspect of bokodes is that distance and orientation of a bokode from a camera can be deduced by processing. The illustration is from the MIT group developing bokodes. Bokode laser pointers. A camera mounted at the front of a room can assign an identifier to all bokodes in range and precisely report where each device is pointing. This means that the class can be issued with cheap pointers, uniquely identifiable, and requiring no batteries. www.bioscience.heacademy.ac.uk/journal/ The late Professor Ed Wood (1941–2008) had many connections with biochemical education and his final editing task has now been published. This is a printed book titled Bioscience Education Collected papers volumes 1–8 and runs to 626 pages. The first edition of Bioscience Education came out in May 2003, when Ed Wood was Director of the Bioscience Education group. For the compilation of volumes 1–8, Ed Wood wrote a preface, in which he emphasized the high standard of editorial review for the journal and how this resulted in a collection of articles that are not ephemeral, but based on underlying fundamentals of teaching. The preface begins “In the last 10–20 years university teaching has undergone a number of profound changes. The universities have not only increased in size (number of students) but also they have increased in number. Furthermore, the electronic revolution began in earnest, leading to the concept of “e-learning” - use of computers, computer-based assessments, asynchronous learning, and an ‘anytime, anyplace’ ethos to course provision. The Internet has pervaded almost all aspects of university life and more recently the Web 2.0 technologies encourage users to alter the content and so become active contributors, rather than passive recipients of information. Such technologies have tremendous potential to improve communication between staff, between staff and students, and between student and student, yet the rapid rate of development means a sense of alienation and being left behind by the ‘digital natives’ is also a real possibility. Additionally, against a background of reducing resource for teaching, increasing class sizes and a much more disparate set of students entering the universities there has been a tremendous emphasis on research activity (almost certainly not pedagogic research in science departments). In the UK this has exacerbated the disparity in recognition, reward and emphasis between research and teaching activities and arguably has rewarded the former much more than the latter.” The book is organized by affinity of content under the section titles; The technology of teaching, Computer based/internet resources, Assessment, Multiple choice questions, Scientific writing/reading, Peer assessment/teaching, Ethics teaching, Laboratory projects, Skills training, and Miscellaneous. The book is free by request to the website given. www.ncbi.nlm.nih.gov/pubmed/19762644 Investigators from a consortium of institutions have constructed a complete model, including three-dimensional protein structures, of the central metabolic pathways of the bacterium Thermotoga maritima. This is the most comprehensive model of metabolism so far produced and is overlaid with atomic resolution structures of the proteins. Analysis of the model shows the important role of a small number of essential protein shapes; providing new insights into the evolution of protein networks (i.e. a single shape can be adapted to perform kindred reactions). The published network has 478 proteins characterized by biochemical studies, structural genomics, and computer modeling. The research may also have applications in energy research because bacteria like T. maritima can be engineered to produce hydrogen as a source of clean energy. The work has been published as Three-dimensional structural view of the central metabolic network of Thermotoga maritima. Science. 2009 Sep 18;325(5947):1544–9. When extended and applied to humans the modeling could help identify drug reactions before clinical trials. www.bdc.ib.unicamp.br/rbebbm/ With 30,000 accesses per week this site is remarkably popular. I was led to it on the commendation of Manuel Costa and others in the BAMBED special edition on biochemical education in Latin America and Portugal (volume 37(5)). At an IUBMB meeting in Salvador Brazil (2007), I saw the enthusiasm of young Brazilians for a career in the molecular life sciences. Brazil, with 192 million people is the fifth most populous country behind China, India, USA, and Indonesia. Most education in Brazil is in the native language of Portuguese, but Brazilians are mindful of teaching and communicating in English to be part of the global community. This website has a wealth of teaching material by way of articles, animations, and teaching aids. The languages are a hybrid of English and Portuguese, often both for a single item. I became a registered user to get access to the free content. I also contributed some of my own material. As Brazil builds its economic standing, we can expect to see biochemical education get ever stronger in Brazil. www.cshperspectives.org Cold Spring Harbor offer several paid subscription titles and have now added a new title Perspectives in Biology that is free online. This journal offers comprehensive surveys of topics in molecular, cell and developmental biology, genetics, neuroscience, immunology, cancer biology, and molecular pathology. Eminent researchers have been commissioned to contribute articles of a high standard that will build to be comprehensive across the target research areas. November 2009 is the fifth of the monthly issues and has feature articles on NF-κB signaling in inflammation and cancer, auxin as a patterning signal for plants and membrane dynamics in cell polarity. Back issue articles are listed in groups by subject content (e.g. Symmetry Breaking in Biology has 18 entries). You can get Twitter or RSS feeds to alert you when new articles are available. The reviews are fulsome and could be used in coursework for student literature studies. www.sloanconsortium.org The Sloan Consortium (USA) supports the integration of online education into the total delivery of higher education. This nonprofit organization does have cost-recovery fees for membership and some services. The consortium particularly aims to facilitate sharing of effective practices to improve online education. At the website, you can find a catalog of degree and certificate programs offered by accredited members. For institutions reviewing their practices the consortium provides speakers and consultants to hosts conferences and workshops. Nevertheless it was the consortium's Journal of Asynchronous Learning Networks that attracted my initial attention. Access to the full journal content requires membership, but there is much to be garnered from abstracts of the articles. The following abstract from the journal is from the article The Road to Retention: A Closer Look at Institutions that Achieve High Course Completion Rates. “From 2003 to 2007, enrollment in online courses in the US nearly doubled, from 1.98 million to 3.94 million. By 2007, the 12.9% growth rate of online enrollments far exceeded the 1.2% growth of the overall higher education student population. Online education will continue to grow as a significant percentage of the 19 million enrollments in postsecondary education projected for 2010. In a climate of shifting demographics and economies, demand for higher education grows. Understanding how students succeed online can lead to far greater access and success, contributing significantly to initiatives for doubling the number of United States college graduates. Yet a barrier to greater access via online education is the belief that online retention rates are lower than in other delivery modes.” For non-members a pdf copy of the article is $5.95. www.youtube.com/education Beginning October 2009 YouTube has presented material from 45 universities in a strand that it has dedicated to education. According to their own promotion, YouTube EDU is a global classroom where everyone can watch and engage with a range of videos that have been uploaded by some of the world's great universities. Harvard is prominently featured. My first impression was that it was a jumble of short news-like bites on subjects of general interest, aimed at browsing visitors. Even pulling up the subcategory of science resulted in choices of mostly short items that focused heavily on the environment. An exception that caught my attention was a lecture submitted by the University of Maryland that took 84 min with questions at the end. The speaker was Steve Fuller, Professor of Sociology at the University of Warwick, UK. The talk revisited C.P. Snow's analysis of the divide between arts and science, sometimes called the science wars. Fuller claims that few recent commentators are familiar with the historical thinking that links Snow's concerns from the mid twentieth century to concerns of the present. It was interesting to me to listen to a speaker reminding me of the many philosophical insights of Snow and numerous other ancient and modern philosophers. After this diversion I found that entering biochemistry into the YouTube search field returned 342 matches. The biochemistry offerings are mostly conventional-length lectures on standard topics in main-stream biochemistry courses. www.ted.com Technology Entertainment And Design (TED) started out in 1984 as a conference to bring together innovators in the three areas of the title. This convergence brings to mind the often quoted phrase attributed to Lenin that “everything is connected to everything”. Whatever we do in education, we are connected with the core interests reflected by TED. The group claim to be dedicated to ideas worth spreading and they offer videos from the famous (e.g. Al Gore and Jane Goodall), the talented and the well qualified. It has been described as YouTube for intellectuals. However, with annual conferences, prizes, and community chats it is much more than that. The home page offers feature talks and you will get extra details by passing your mouse over each panel. There is a section on science and be sure to click on view all tags after browsing the default feature talks. I was well rewarded when I chose a presentation by Jonathan Drori titled “Why we're storing billions of seeds”. I encountered only excellence in my sampling of the big and small issues that are covered by TED. research.microsoft.com/apps/pubs/default.aspx?id=76529 This site leads to a article titled Cyberchondria: Studies of the Escalation of Medical Concerns in Web Search, authored by Ryen White and Eric Horvitz who are research staff employed by Microsoft. Cyberchondria is a term coined to describe hypochondriac behavior generated by undiscerning use of medical data obtained from the web. While the term cyberchondria is interesting in itself, the bigger-picture problem this represents is that of students believing flawed information from the web. The assertion that branched chain amino acids promote athletic strength is one of the flawed “facts” that comes to mind (it is not supported by reputable studies). As teachers we need to give students the skill to develop a baloney-meter (Carl Sagan's phrase) to appreciate credible and relevant information from the web, or wherever it is sourced. A way of looking at cyberchondria is as a form of medical-student-disease, which applies when medical students suffer the symptoms of the disease they are studying. A headache is unlikely to be viral encephalitis, but that does not stop a hypochondriac from worrying that it is the cause. The research for this article on cyberchondria involved a survey of 515 individuals who made health-related searches. Susceptible individuals suffered high anxiety and disruption of their work. Because the writers are employed by Microsoft the article has a focus that is not so much to fix the cyberchondriac, as to fix web-searching protocols and the structure of the content to avoid cyberchondria. The concluding comments to the article are “Search engine architects have a responsibility to ensure that searchers do not experience unnecessary concern generated by the definitions of relevance and the ranking algorithms their engines use. They must be cognizant of the potential challenges of cyberchondria, and focus on serving medical search results that are reliable, complete, and timely, as well as topically relevant. Directly tackling cyberchondria is an opportunity to leverage readily available expertise in the information-retrieval and medical informatics communities in areas, such as document ranking, user modeling, machine learning, and user interface design for the direct benefit of the many people turning to the Web to interpret common medical symptoms.”

Knowledge of the preferences of individual users is essential for intelligent systems whose performance is tailored for individual users, such as agents that interact with human users, instructional environments, and learning apprentice systems. Various memory-based, instance-based, and case-based systems have been developed for preference modeling, but these system have generally not addressed the task of selecting examples to use as queries to the user. This paper describes UGAMA, an approach to learning preference criteria through active exploration. Under this approach, Unit Gradient Approximations (UGAs) of the underlying quality function are obtained at a set of reference points through a series of queries to the user. Equivalence sets of UGAs are then merged and aligned (MA) with the apparent boundaries between linear regions. In an empirical evaluation with artificial data, use of UGAs as training data for an instance-based ranking algorithm (1ARC) led to more accurate ranking than training with random instances, and use of UGAMA led to greater ranking accuracy than UGAs alone.

One related area that has received little attention with regards to AODE is the use of attribute weights for ranking. This paper investigates how to learn an AWAODE with accurate ranking from data sets. We first explore various methods, such as gain ratio, correlation-based feature selection attribute selection algorithm, mutual information and relief attribute ranking algorithm. Our experiments clearly show that an attribute weighted AODE trained to produce AUC ranking outperforms AODE and NB. Then, we propose a new approach to weight AODE for generating accurate ranking, called decision tree-based attribute weighted averaged one-dependence estimator, simply DTWAODE. In DTWAODE, the weight for an attribute is set according to its depth in the decision tree building on the training samples. The experimental results show that our new attribute weighted model via AODE performance effectively than AODE and other attribute weighted approaches on AUC.

The popularity of Web 2.0 applications promotes the emergence of user generated content (UGC), e.g., the comments in blogsphere, and the UGC reflects the viewpoints of web users towards a specific event or product. In this paper, we propose a summarization model which applies a novel sentence similarity measure. In the proposed two-stage similarity measure, we utilize a learning method based on an optimization perspective to combine different types of similarity for a refined similarity measure. Both standard cosine similarity and topic based similarity measure are explored to compute the preliminary similarity. In order to evaluate the novel similarity measure, we conduct experiments on a real-world blog data set and the result proves the effectiveness of our proposed method.KeywordsComments-oriented SummarizationBlogsphereSentence Similarity

Improved search quality enhances users’ satisfaction, which directly impacts sales growth of an E-Commerce (E-Com) platform. Traditional Learning to Rank (LTR) algorithms require relevance judgments on products. In E-Com, getting such judgments poses an immense challenge. In the literature, it is proposed to employ user feedback (such as clicks, add-to-basket (AtB) clicks and orders) to generate relevance judgments. It is done in two steps: first, query-product pair data are aggregated from the logs and then order rate etc. are calculated for each pair in the logs. In this paper, we advocate counterfactual risk minimization (CRM) approach which circumvents the need of relevance judgements, data aggregation and is better suited for learning from logged data, i.e. contextual bandit feedback. Due to unavailability of public E-Com LTR dataset, we provide Mercateo dataset from our platform. It contains more than 10 million AtB click logs and 1 million order logs from a catalogue of about 3.5 million products associated with 3060 queries. To the best of our knowledge, this is the first work which examines effectiveness of CRM approach in learning ranking model from real-world logged data. Our empirical evaluation shows that our CRM approach learns effectively from logged data and beats a strong baseline ranker (\(\lambda \)-MART) by a huge margin. Our method outperforms full-information loss (e.g. cross-entropy) on various deep neural network models. These findings demonstrate that by adopting CRM approach, E-Com platforms can get better product search quality compared to full-information approach.

Evolutionary information is the primary tool for detecting functional conservation in nucleic acid and protein. This information has been extensively used to predict structure, interactions and functions in macromolecules. Pathogenicity prediction models rely on multiple sequence alignment information at different levels. However, most accurate genome-wide variant deleteriousness ranking algorithms consider different features to assess the impact of variants. Here, we analyze three different ways of extracting evolutionary information from sequence alignments in the context of pathogenicity predictions at DNA and protein levels. We showed that protein sequence-based information is slightly more informative in the annotation of Clinvar missense variants than those obtained at the DNA level. Furthermore, to achieve the performance of state-of-the-art methods, such as CADD and REVEL, the conservation of reference and variant, encoded as frequencies of reference/alternate alleles or wild-type/mutant residues, should be included. Our results on a large set of missense variants show that a basic method based on three input features derived from the protein sequence profile performs similarly to the CADD algorithm which uses hundreds of genomic features. As expected, our method results in ~ 3% lower area under the receiver-operating characteristic curve (AUC). When compared with an ensemble-based algorithm (REVEL). Nevertheless, the combination of predictions of multiple methods can help to identify more reliable predictions. These observations indicate that for missense variants, evolutionary information, when properly encoded, plays the primary role in ranking pathogenicity.

This paper presents our experimental work on mining of opinions from a large number of blog posts and its relevance for socio-political research. The experimental work involves collecting blog data on three interesting topics, transforming the collected blog data into vector space representation, and then performing opinion mining using both a machine learning text classifier and an unsupervised semantic orientation approach. We implemented Naïve Bayes and SO-PMI-IR algorithms for opinion mining. We obtained interesting results, which have been evaluated for correctness and also cross-validated with the outcomes of multiple techniques employed. The paper concludes with a short discussion of the results and relevance of the experimental work.

This paper proposes the CorpRank algorithm to extract social hierarchies from electronic communication data. The algorithm computes a ranking score for each user as a weighted combination of the number of emails, the number of responses, average response time, clique scores, and several degree and centrality measures. The algorithm uses principal component analysis to calculate the weights of the features. This score ranks users according to their importance, and its output is used to reconstruct an organization chart. We illustrate the algorithm over real‐world data using the Enron corporation’s e‐mail archive. Compared to the actual corporate work chart, compensation lists, judicial proceedings, and analyzing the major players involved, the results show promise.

Finding a meaningful and fulfilling work and finding right talent for a given job is a challenging and classical Human Capital Management (HCM) problem. For a long time, HCM was carried out in an offline manner requiring human interventions. As scale of organizations have grown, it becomes relevant to automate these processes. In this paper, we present and detail a stateless scalable architecture for an automated HCM system. For clustering and categorizing job postings and candidate profiles we present algorithms that uses Machine Learning techniques. We present an algorithm that uses Natural Language Processing for feature extraction. We also present a ranking algorithm that uses semantic web technologies for providing accurate recommendations. We conclude this paper presenting results of presented algorithms on real-world job posting and candidate data.

In order to choose from a list of functionally similar services, users often need to make their decisions based on multiple QoS criteria they require on the target service. In this process, different users may follow different decision making strategies, some are compensatory and some are non-compensatory. Most of the current QoS-based service selection systems do not consider these decision strategies in the ranking process, which we believe are crucial for generating accurate ranking results for individual users. In this thesis, we propose a decision strategy based service ranking model. Furthermore, considering that different users follow different strategies in different contexts at different times, we apply a learning to rank algorithm to learn a personalized ranking model for individual users based on how they select services in the past. Our experiment result shows the effectiveness of the proposed approach.