Using Dynamic Multi-Task Non-Negative Matrix Factorization to Detect the Evolution of User Preferences in Collaborative Filtering

Predicting what items will be selected by a target user in the future is an important function for recommendation systems. Matrix factorization techniques have been shown to achieve good performance on temporal rating-type data, but little is known about temporal item selection data. In this paper, we developed a unified model that combines Multi-task Non-negative Matrix Factorization and Linear Dynamical Systems to capture the evolution of user preferences. Specifically, user and item features are projected into latent factor space by factoring co-occurrence matrices into a common basis item-factor matrix and multiple factor-user matrices. Moreover, we represented both within and between relationships of multiple factor-user matrices using a state transition matrix to capture the changes in user preferences over time. The experiments show that our proposed algorithm outperforms the other algorithms on two real datasets, which were extracted from Netflix movies and Last.fm music. Furthermore, our model provides a novel dynamic topic model for tracking the evolution of the behavior of a user over time.     

Associations between Film Preferences and Risk Factors for Suicide: An Online Survey

Several studies indicate that exposure to suicide in movies is linked to subsequent imitative suicidal behavior, so-called copycat suicides, but little is currently known about whether the link between exposure to suicidal movies and suicidality is reflected in individual film preferences. 943 individuals participated in an online survey. We assessed associations between preferred film genres as well as individual exposure to and rating of 50 pre-selected films (including 25 featuring a suicide) with suicidal ideation, hopelessness, depression, life satisfaction, and psychoticism. Multiple regression analyses showed that preferences for film noir movies and milieu dramas were associated with higher scores on suicidal ideation, depression and psychoticism, and low scores on life satisfaction. Furthermore, preferences for thrillers and horror movies as well as preferences for tragicomedies, tragedies and melodramas were associated with higher scores of some of the suicide risk factors. There was also a dose-response relationship between positive rating of suicide films and higher life satisfaction. Due to the cross-sectional design of the study causality cannot be assessed. Individual film genre preferences seem to reflect risk factors of suicide, with film genres focusing on sad contents being preferred by individuals with higher scores on suicide risk factors. However, suicide movies are more enjoyed by viewers with higher life satisfaction, which may reflect a better ability to cope with such content.     

AUI&GIV: Recommendation with Asymmetric User Influence and Global Importance Value

The user-based collaborative filtering (CF) algorithm is one of the most popular approaches for making recommendation. Despite its success, the traditional user-based CF algorithm suffers one serious problem that it only measures the influence between two users based on their symmetric similarities calculated by their consumption histories. It means that, for a pair of users, the influences on each other are the same, which however may not be true. Intuitively, an expert may have an impact on a novice user but a novice user may not affect an expert at all. Besides, each user may possess a global importance factor that affects his/her influence to the remaining users. To this end, in this paper, we propose an asymmetric user influence model to measure the directed influence between two users and adopt the PageRank algorithm to calculate the global importance value of each user. And then the directed influence values and the global importance values are integrated to deduce the final influence values between two users. Finally, we use the final influence values to improve the performance of the traditional user-based CF algorithm. Extensive experiments have been conducted, the results of which have confirmed that both the asymmetric user influence model and global importance value play key roles in improving recommendation accuracy, and hence the proposed method significantly outperforms the existing recommendation algorithms, in particular the user-based CF algorithm on the datasets of high rating density.     

N- and C-capping preferences for all 20 amino acids in alpha-helical peptides.

We have determined the N- and C-capping preferences of all 20 amino acids by substituting residue X in the peptides NH2-XAKAAAAKAAAAKAAGY-CONH2 and in Ac-YGAAKAAAAKAAAAKAX-CO2H. Helix contents were measured by CD spectroscopy to obtain rank orders of capping preferences. The data were further analyzed by our modified Lifson-Roig helix-coil theory, which includes capping parameters (n and c), to find free energies of capping (-RT ln n and -RT ln c), relative to Ala. Results were obtained for charged and uncharged termini and for different charged states of titratable side chains. N-cap preferences varied from Asn (best) to Gln (worst). We find, as expected, that amino acids that can accept hydrogen bonds from otherwise free backbone NH groups, such as Asn, Asp, Ser, Thr, and Cys generally have the highest N-cap preference. Gly and acetyl group are favored, as are negative charges in side chains and at the N-terminus. Our N-cap preference scale agrees well with preferences in proteins. In contrast, we find little variation when changing the identity of the C-cap residue. We find no preference for Gly at the C-cap in contrast to the situation in proteins. Both N-cap and C-cap results for Tyr and Trp are inaccurate because their aromatic groups affect the CD spectrum. The data presented here are of value in rationalizing mutations at capping sites in proteins and in predicting the helix contents of peptides.     

Information Filtering via Heterogeneous Diffusion in Online Bipartite Networks

The rapid expansion of Internet brings us overwhelming online information, which is impossible for an individual to go through all of it. Therefore, recommender systems were created to help people dig through this abundance of information. In networks composed by users and objects, recommender algorithms based on diffusion have been proven to be one of the best performing methods. Previous works considered the diffusion process from user to object, and from object to user to be equivalent. We show in this work that it is not the case and we improve the quality of the recommendation by taking into account the asymmetrical nature of this process. We apply this idea to modify the state-of-the-art recommendation methods. The simulation results show that the new methods can outperform these existing methods in both recommendation accuracy and diversity. Finally, this modification is checked to be able to improve the recommendation in a realistic case.     

Item Anomaly Detection Based on Dynamic Partition for Time Series in Recommender Systems

In recent years, recommender systems have become an effective method to process information overload. However, recommendation technology still suffers from many problems. One of the problems is shilling attacks-attackers inject spam user profiles to disturb the list of recommendation items. There are two characteristics of all types of shilling attacks: 1) Item abnormality: The rating of target items is always maximum or minimum; and 2) Attack promptness: It takes only a very short period time to inject attack profiles. Some papers have proposed item anomaly detection methods based on these two characteristics, but their detection rate, false alarm rate, and universality need to be further improved. To solve these problems, this paper proposes an item anomaly detection method based on dynamic partitioning for time series. This method first dynamically partitions item-rating time series based on important points. Then, we use chi square distribution (χ²) to detect abnormal intervals. The experimental results on MovieLens 100K and 1M indicate that this approach has a high detection rate and a low false alarm rate and is stable toward different attack models and filler sizes.     

Some like it hot: body and weapon size affect thermoregulation in horned beetles

Body size and shape affect thermoregulatory properties of organisms, and in turn are believed to have shaped macroevolutionary patterns of morphological diversity across many taxa. However, it is less clear whether thermoregulation plays a role in shaping intraspecific morphological diversity such as sexual dimorphisms or the conditional expression of exaggerated secondary sexual traits. Here, we investigate individual thermoregulatory properties in two species of horned beetles that share similar ecologies and body size ranges, but differ substantially in degree of sexual and male dimorphism. We find that intraspecific variation in body size had an unexpectedly large effect on thermal preference behavior and the ability to passively regulate body temperature. Furthermore, we find that the presence or absence of exaggerated secondary sexual traits dramatically altered thermal preference behavior, consistent with a thermoregulatory cost of horn possession. Lastly, we show that the increase in surface area associated with the expression of enlarged horns is, by itself, insufficient to account for the radically altered thermoregulatory behavior observed in horn-bearing males, and discuss possible alternative, physiological explanations. These findings are among the first to link intra-and interspecific variation in body- and weapon size to thermal preferences within and between insect species.     

When flexibility is stable: implicit long-term shaping of olfactory preferences

Preferences are traditionally assumed to be stable. However, empirical evidence such as preference modulation following choices calls this assumption into question. The evolution of such postchoice preference over long time spans, even when choices have been explicitly forgotten, has so far not been studied. In two experiments, we investigated this question by using a variant of the free choice paradigm: In a first session, participants evaluated the pleasantness of a number of odors. We then formed pairs of similarly rated odors, and asked participants to choose their favorite, for each pair. Participants were then presented with all odors again, and asked for another pleasantness rating. In a second session 1 week later, a third pleasantness rating was obtained, and participants were again asked to choose between the same options. Results suggested postchoice preference modulation immediately and 1 week after choice for both chosen and rejected options, even when choices were not explicitly remembered. A third experiment, using another paradigm, confirmed that choice can have a modulatory impact on preferences, and that this modulation can be long-lasting. Taken together, these findings suggest that although preferences appear to be flexible because they are modulated by choices, this modulation also appears to be stable over time and even without explicit recollection of the choice. These results bring a new argument to the idea that postchoice preference modulation could rely on implicit mechanisms, and are consistent with the recent proposal that cognitive dissonance reduction could to some extent be implicit.     

Preferences for masculinity in male bodies change across the menstrual cycle

In human females cyclic shifts in preference have been documented for odour and physical and behavioral male traits. Women prefer the smell of dominant males, more masculine male faces and men behaving more dominantly when at peak fertility than at other times in their menstrual cycle. Here we examine variation in preferences for body sexual dimorphism. Across two studies, both between- and within-participant, we show that women prefer greater masculinity in male bodies at times when their fertility is likely highest, in the follicular phase of their cycle. Shifts were seen when rating for a short-term but not when rating for a long-term relationship. In line with studies showing similar effects for facial sexual dimorphism, we also show that women prefer greater masculinity when they think themselves attractive than when they think themselves less attractive. These results indicate that women's preferences for sexual dimorphism in male bodies follow a similar pattern as found for sexual dimorphism and dominance in other domains and such differences in preference may serve a similar function. Cyclic preferences could influence women to select partners when most likely to become pregnant that possess traits that may be most likely to maximize their offspring's quality via attraction to masculinity or serve to help acquire investment via attraction to femininity.     

A cognitive framework for mate choice and species recognition

Mating decisions contribute to both the fitness of individuals and the emergence of evolutionary diversity, yet little is known about their cognitive architecture. We propose a simple model that describes how preferences are translated into decisions and how seemingly disparate patterns of preference can emerge from a single perceptual process. The model proposes that females use error-prone estimates of attractiveness to select mates based on a simple decision rule: choose the most attractive available male that exceeds some minimal criterion. We test the model in the tungara frog, a well-characterized species with an apparent dissociation between mechanisms of mate choice and species recognition. As suggested by our model results, we find that a mate attraction feature alters assessments of species status. Next, we compare female preferences in one-choice and two-choice tests, contexts thought to emphasize species recognition and mate choice, respectively. To do so, we use the model to generate maximum-likelihood estimators of preference strengths from empirical data. We find that a single representation of preferences is sufficient to explain response probabilities in both contexts across a wide range of stimuli. In this species, mate choice and species recognition are accurately and simply summarized by our model. While the findings resolve long-standing anomalies, they also illustrate how models of choice can bridge theoretical and empirical treatments of animal decisions. The data demonstrate a remarkable congruity of perceptual processes across contexts, tasks, and taxa.     

THE RELIABILITY OF MDPREF TO SHOW INDIVIDUAL PREFERENCE

Liking rating data of 8 pigmeat products from 99 Australian females aged 25-45 was analyzed using MDPREF. A line scale extended past the anchor points, 100 mm long and anchored by the words 'dislike extremely' and 'like extremely' was used. Liking was judged on the product name alone, no products were actually tasted. MDPREF configures products and consumers so that individual preferences (or product order preference) can be obtained. The order of product preference from the rating scores and order of product preference from the MDPREF analysis for each consumer was matched. It was found that the preference order for each consumer from the MDPREF map matched the order shown by rating scores in about 25% of cases. MDPREF produced 17% of consumers who had no products in the correct position for the preference order. Only 1% of consumers had all eight product preferences in the correct order.     

Stochasticity in Sexual Selection Enables Divergence: Implications for Moth Pheromone Evolution

Sexual selection has long been hypothesized to lead to allopatric speciation, and one possible mechanism for this is that its interaction with stochasticity, which perturbs the trait and preference equilibria, can result in different traits being preferred in different populations. Here we specifically examine the role that stochastic changes in sexual selection strength plays in the shift of predominance between pairs of preferences and traits within a single population. We first create a single-locus null model of stochasticity during frequency dependent selection and then compare it to a two-locus population genetic model with stochastic strengths of female preferences for male traits. We find some interesting differences between the two models, primarily that in the two-locus sexual selection model shifts between preference and trait regimes occur more often with both weak and strong preferences, compared to intermediate preference strengths. We discuss the implications of our results for the evolution of pheromone blends and male responses during speciation in moths, a case that seems to match the assumptions of our model.     

Group Colocation Behavior in Technological Social Networks

We analyze two large datasets from technological networks with location and social data: user location records from an online location-based social networking service, and anonymized telecommunications data from a European cellphone operator, in order to investigate the differences between individual and group behavior with respect to physical location. We discover agreements between the two datasets: firstly, that individuals are more likely to meet with one friend at a place they have not visited before, but tend to meet at familiar locations when with a larger group. We also find that groups of individuals are more likely to meet at places that their other friends have visited, and that the type of a place strongly affects the propensity for groups to meet there. These differences between group and solo mobility has potential technological applications, for example, in venue recommendation in location-based social networks.     

Assessment of landscape aesthetics—Validation of a landscape metrics-based assessment by visual estimation of the scenic beauty

The assessment of cultural ecosystem services, in our case landscape aesthetics, is the most commonly investigated but least formalized issue in the scope of the ecosystem services concept. In contrast to ecological or economic aspects, the assessment of aesthetics cannot easily be based on quantitative information. Therefore, two different methodological approaches that assess landscape aesthetics either from an objective or a subjective point of view have been established in the past.This article presents in its first part an objective, landscape metrics-based assessment approach. We defined naturalness and landscape diversity as assessment criteria and selected Shannon's Diversity Index (SHDI), Shape Index (SHAPE) and Patch Density (PD) as indicators. We tested our approach for a set of nine different landscape types in a model region in Saxony, Germany.For validating the developed methodology, we carried out a survey with 153 participants in order to investigate their subjective preferences for the different landscape types. These preferences had to be expressed by rating the landscape types on a scale from 1 (very ugly) to 5 (very beautiful). The study was based on three different data sets, namely photographs of the landscape types, satellite images, and land cover maps.Statistical tests were applied (a) to investigate the impact of personal factors on the ratings, (b) to detect whether abstraction levels are suitable for preference studies, and (c) to compare the results of the objective approach (landscape metrics) and the subjective approach (visual assessment). Personal factors did not influence the visual assessment results significantly. We found the highest correlation of the landscape metrics-based assessment with the visual assessment results of the photographs. We conclude that the three landscape metrics might be applied to the monitoring of landscape aesthetics. An extended study with more participants might be useful to further investigate the reliability of our findings.     

The effects of sexual selection on trait divergence in a peripheral population with gene flow

The unique aspects of speciation and divergence in peripheral populations have long sparked much research. Unidirectional migration, received by some peripheral populations, can hinder the evolution of distinct differences from their founding populations. Here, we explore the effects that sexual selection, long hypothesized to drive the divergence of distinct traits used in mate choice, can play in the evolution of such traits in a partially isolated peripheral population. Using population genetic continent‐island models, we show that with phenotype matching, sexual selection increases the frequency of an island‐specific mating trait only when female preferences are of intermediate strength. We identify regions of preference strength for which sexual selection can instead cause an island‐specific trait to be lost, even when it would have otherwise been maintained at migration‐selection balance. When there are instead separate preference and trait loci, we find that sexual selection can lead to low trait frequencies or trait loss when female preferences are weak to intermediate, but that sexual selection can increase trait frequencies when preferences are strong. We also show that novel preference strengths almost universally cannot increase, under either mating mechanism, precluding the evolution of premating isolation in peripheral populations at the early stages of species divergence.     

Optimizing Online Social Networks for Information Propagation

Online users nowadays are facing serious information overload problem. In recent years, recommender systems have been widely studied to help people find relevant information. Adaptive social recommendation is one of these systems in which the connections in the online social networks are optimized for the information propagation so that users can receive interesting news or stories from their leaders. Validation of such adaptive social recommendation methods in the literature assumes uniform distribution of users'' activity frequency. In this paper, our empirical analysis shows that the distribution of online users'' activity is actually heterogenous. Accordingly, we propose a more realistic multi-agent model in which users'' activity frequency are drawn from a power-law distribution. We find that previous social recommendation methods lead to serious delay of information propagation since many users are connected to inactive leaders. To solve this problem, we design a new similarity measure which takes into account users'' activity frequencies. With this similarity measure, the average delay is significantly shortened and the recommendation accuracy is largely improved.     

Mining-based associative image filtering using harmonic mean

With the development of IT convergence technologies, users can now more easily access useful information. These days, diverse and far-reaching information is being rapidly produced and distributed instantly in digitized format. Studies are continuously seeking to develop more efficient methods of delivering information to a greater number of users. Image filtering, which extracts features of interest from images, was developed to address the weakness of collaborative filtering, which is limited to superficial data analysis. However, image filtering has its own weakness of requiring complicated calculations to obtain the similarity between images. In this study, to resolve these problems, we propose associative image filtering based on the mining method utilizing the harmonic mean. Using data mining’s Apriori algorithm, this study investigated the association among preferred images from an associative image group and obtained a prediction based on user preference mean. In so doing, we observed a positive relationship between the various image preferences and the various distances between images’ color histograms. Preference mean was calculated based on the arithmetic mean, geometric mean, and harmonic mean. We found through performance analysis that the harmonic mean had the highest accuracy. In associative image filtering, we used the harmonic mean in order to anticipate preferences. In testing accuracy with MAE utilizing the proposed method, this study demonstrated an improvement of approximately 12 % on average compared to previous collaborative image filtering.     

A context-aware personalized resource recommendation for pervasive learning

As it is difficult for learners to discover and obtain the most appropriate resources from massive education resources according to traditional keyword searching method, the context-aware based resource recommendation service becomes a significant part of pervasive learning environments. At present, recommendation mechanisms are widely used in e-commerce field, where content-based or collaborative-based filter strategies are usually considered separately. However, in these existing recommendation mechanisms, the dynamic interests and preference of learners, the access pattern and the other attributes of pervasive learning environments (such as multi-modes connecting and resources distribution) are always neglected. Thus, these mechanisms can not effectively reflect learners’ actual preference and can not adapt to pervasive learning environments perfectly. To address these problems, a context-aware resource recommendation model and relevant recommendation algorithm for pervasive learning environments are proposed. Therein, with taking into account the relevant contextual information, the calculation of relevant degree between learners and resources can be divided into two main parts: logic-based RRD (resource relevant degree) and situation-based RRD. In the first part, content-based and collaborative-based recommendation mechanisms are combined together, where the individual preference tree (IPT) is introduced to take into account the multi-dimensional attributes of resources, learners’ rating matrix and the energy of access preference. Meanwhile, the learners’ historical sequential patterns of resource accessing are also considered to further improve the accuracy of recommendation. In the second part, in order to enhance the validation of recommendation, the connecting type relevance and time satisfaction degree are calculated according to other relevant contexts. Then, the candidate resources can be filtered and sorted via combining these two parts to generate (Top-N) recommendation results. The simulations show that our newly proposed method outperforms other state of-the-art algorithms on traditional and newly presented metrics and it may also be more suitable for pervasive learning environments. Finally, a prototype system is implemented based on SEU-ESP to demonstrate the relevant recommendation process further.     

The Power of Ground User in Recommender Systems

Accuracy and diversity are two important aspects to evaluate the performance of recommender systems. Two diffusion-based methods were proposed respectively inspired by the mass diffusion (MD) and heat conduction (HC) processes on networks. It has been pointed out that MD has high recommendation accuracy yet low diversity, while HC succeeds in seeking out novel or niche items but with relatively low accuracy. The accuracy-diversity dilemma is a long-term challenge in recommender systems. To solve this problem, we introduced a background temperature by adding a ground user who connects to all the items in the user-item bipartite network. Performing the HC algorithm on the network with ground user (GHC), it showed that the accuracy can be largely improved while keeping the diversity. Furthermore, we proposed a weighted form of the ground user (WGHC) by assigning some weights to the newly added links between the ground user and the items. By turning the weight as a free parameter, an optimal value subject to the highest accuracy is obtained. Experimental results on three benchmark data sets showed that the WGHC outperforms the state-of-the-art method MD for both accuracy and diversity.