Preferential text classification: learning algorithms and evaluation measures

In many applicative contexts in which textual documents are labelled with thematic categories, a distinction is made between the primary categories of a document, which represent the topics that are central to it, and its secondary categories, which represent topics that the document only touches upon. We contend that this distinction, so far neglected in text categorization research, is important and deserves to be explicitly tackled. The contribution of this paper is threefold. First, we propose an evaluation measure for this preferential text categorization task, whereby different kinds of misclassifications involving either primary or secondary categories have a different impact on effectiveness. Second, we establish several baseline results for this task on a well-known benchmark for patent classification in which the distinction between primary and secondary categories is present; these results are obtained by reformulating the preferential text categorization task in terms of well established classification problems, such as single and/or multi-label multiclass classification; state-of-the-art learning technology such as SVMs and kernel-based methods are used. Third, we improve on these results by using a recently proposed class of algorithms explicitly devised for learning from training data expressed in preferential form, i.e., in the form “for document d _i , category c′ is preferred to category c′′”; this allows us to distinguish between primary and secondary categories not only in the classification phase but also in the learning phase, thus differentiating their impact on the classifiers to be generated.     

A virtual experience to strengthen students' understanding of inference in experimental design

Although our students correctly define the terms p‐value, Type I and Type II errors, they sometimes misinterpret results from real data. In this work we present an assignment intended to clear up these misconceptions.     

Commercializing the laboratory: Faculty patenting and the open science environment

This paper investigates the relationship between patenting and publication of research results by university faculty members. Our study adds to the limited evidence on this topic with an empirical investigation based on a panel data set for a broad sample of university researchers. Results suggest that publication and patenting are complementary, not substitute, activities for faculty members. This is not consistent with recent concerns regarding deleterious effects of patenting on the research output of faculty members. Average citations to publications, however, appear to decline for repeat patenters, suggesting either a decrease in quality or restrictions on use associated in patent protection.     

Nearest neighbor search in metric spaces through Content-Addressable Networks

Most of the Peer-to-Peer search techniques proposed in the recent years have focused on the single-key retrieval. However, similarity search in metric spaces represents an important paradigm for content-based retrieval in many applications. In this paper we introduce an extension of the well-known Content-Addressable Network paradigm to support storage and retrieval of more generic metric space objects. In particular we address the problem of executing the nearest neighbors queries, and propose three different algorithms of query execution. An extensive experimental study on real-life data sets explores the performance characteristics of the proposed algorithms by showing their advantages and disadvantages.     

Effects of fast-velocity eccentric resistance training on early and late rate of force development

This study examined whether short-term maximal resistance training employing fast-velocity eccentric knee extensor actions would induce improvements in maximal isometric torque and rate of force development (RFD) at early (<100 ms) and late phases (>100 ms) of rising torque. Twenty healthy men were assigned to two experimental groups: eccentric resistance training (TG) or control (CG). Participants on the TG trained three days a week for a total of eight weeks. Training consisted of maximal unilateral eccentric knee extensors actions performed at 180°s-1. Maximal isometric knee extensor torque (MVC) and incremental RFD in successive 50 ms time-windows from the onset contraction were analysed in absolute terms (RFD_INC) or when normalised relative to MVC (RFD_REL). After eight weeks, TG demonstrated increases in MVC (28%), RFD_INC (0–50 ms: 30%; 50–100 ms: 31%) and RFD_REL (0–50 ms: 29%; 50–100 ms: 32%). Moreover, no changes in the late phase of incremental RFD were observed in TG. No changes were found in the CG. In summary, we have demonstrated, in active individuals, that a short period of resistance training performed with eccentric fast-velocity isokinetic muscle contractions is able to enhance RFD_INC and RFD_REL obtained at the early phase of rising joint torque.     

Generating suggestions for queries in the long tail with an inverted index

This paper proposes an efficient and effective solution to the problem of choosing the queries to suggest to web search engine users in order to help them in rapidly satisfying their information needs. By exploiting a weak function for assessing the similarity between the current query and the knowledge base built from historical users’ sessions, we re-conduct the suggestion generation phase to the processing of a full-text query over an inverted index. The resulting query recommendation technique is very efficient and scalable, and is less affected by the data-sparsity problem than most state-of-the-art proposals. Thus, it is particularly effective in generating suggestions for rare queries occurring in the long tail of the query popularity distribution. The quality of suggestions generated is assessed by evaluating the effectiveness in forecasting the users’ behavior recorded in historical query logs, and on the basis of the results of a reproducible user study conducted on publicly-available, human-assessed data. The experimental evaluation conducted shows that our proposal remarkably outperforms two other state-of-the-art solutions, and that it can generate useful suggestions even for rare and never seen queries.     

“Too complex for me!” Why do performance-approach and performance-avoidance goals predict exam performance?

Classroom research on achievement goals has revealed that performance-approach goals (goals to outperform others) positively predict exam performance whereas performance-avoidance goals (goals not to perform more poorly than others) negatively predict it. Because prior classroom research has primarily utilized multiple-choice exam performance, the first aim of the present study was to extend these findings to a different measure of exam performance (oral examination). The second aim of this research was to test the mediating role of perceived difficulty. Participants were 49 4th year psychology students of the University of Geneva. Participants answered a questionnaire assessing their level of performance-approach and performance-avoidance goal endorsement in one of their classes as well as the perceived difficulty of this class for themselves. Results indicated that performance-approach goals significantly and positively predicted exam grades. Performance-avoidance goals significantly and negatively predicted grades. Both of these relationships were mediated by the perceived difficulty of the class for oneself. Thus, the links previously observed between performance goals and exam performance were replicated on an oral exam. Perceived difficulty is discussed as a key dimension responsible for these findings.     

Nearest neighbor search in metric spaces through Content-Addressable Networks

Most of the peer-to-peer search techniques proposed in the recent years have focused on the single-key retrieval. However, similarity search in metric spaces represents an important paradigm for content-based retrieval in many applications. In this paper we introduce an extension of the well-known Content-Addressable Network paradigm to support storage and retrieval of more generic metric space objects. In particular we address the problem of executing the nearest neighbors queries, and propose three different algorithms of query propagation. An extensive experimental study on real-life data sets explores the performance characteristics of the proposed algorithms by showing their advantages and disadvantages.     

Reliability,usefulness and construct validity of the Combined Basketball Skill Test (CBST)

This study aimed to assess the reliability, usefulness and construct validity of the newly developed Combined Basketball Skill Test (CBST). Fifteen recreational (age = 22.8 ± 4.2 y, stature = 184.8 ± 6.5 cm, body mass = 81.6 ± 9.6 kg, training experience = 9.8 ± 5.3 y) and fifteen semiprofessional (age = 18.9 ± 3.3 y, stature = 190.5 ± 8.1 cm, body mass = 84.2 ± 11.2 kg, training experience = 11.1 ± 3.5 y) players volunteered to participate in this study. Test–retest reliability and usefulness were examined for recreational players, while construct validity was evaluated comparing the two player groups. The CBST is composed of 12 trials and its outcome measures include: completion time (sum of the 12 trial times); penalty time (sum of the times from the 12 trials); performance time (completion time + penalty time) and total number of errors. Relative reliability analysis showed acceptable ICC values (i.e. ≥0.70) in all the studied variables. Absolute reliability analysis showed a CV < 5% for completion (1.6%) and performance (2.0%) time, while a CV >5% is reported for the remaining variables. The usefulness of the test was considered “Marginal” and “Good” when comparing TE values with SWC₀₂ and SWC₀₅, respectively for all the studied variables. Likely and very likely differences were shown between recreational and semiprofessional players in all investigated variables. Results showed that the CBST is reliable, useful to detect moderate changes and valid to assess basketball skills.     

Between al-Andalus and Genoa: The “Embriaci Tower” Lustreware

This article examines a set of lustreware pottery discovered in the fill of the tower belonging to the Genoese family, the Embriaci. The finds are datable to the mid-thirteenth century, characterised by decorative patterns which are atypical in lustreware production in al-Andalus, though some comparisons can be drawn with the decorative patterns of the “vasos de la Alhambra” and the pottery known as “nazarí primitive”. The context of the discovery in Genoa will briefly be outlined, including the story of the family settlement (Part 1), the stratigraphic sequence (Part 2), and the stylistic connections to other pottery characterising it (Part 3). The formal and decorative elements characterising the lustreware pottery found in the Embriaci Tower will then be examined and compared (Part 4) with finds from the Upper Tyrrhenian region (Provence, Savona, Pisa). Part 5 will suggest that the finds are a specific type of decorated lustreware produced in the thirteenth century in al-Andalus (Murcia, Málaga and Almería). The concluding part (Part 6) will investigate the meaning of the presence of the set of imported ceramics in the “curia” of the Embriaci family in Genoa.

This type of pottery has only been discovered hitherto in excavation finds and architectural contexts (the so called “bacini”) in Italy and France. The identification of decorative and formal elements characterising this “group” of pottery will allow the identification of examples of this production outside of Italy and France. The publication of the findings will provide new data to develop an archaeological overview of commercial contacts between al-Andalus and the Tyrrhenian area (Provence, Liguria, Tuscany) during the thirteenth century.