Molecular clustering in the liquid state is controlled by the relativity of interaction energies. This means that the clustering is strongly promoted by the coexistence of relatively weakly interacting molecules in the solution. This relativity-controlled clustering was observed for 1-pentanol clustering through the mass spectrometry for clusters isolated from liquid droplets via adiabatic expansion in a vacuum chamber. The 1-pentanol clustering through hydrogen-bonding interaction was significantly promoted by the mixing with water, methanol, acetonitrile or dichloromethane, but it was not promoted by the mixing with 1-propanol or 1, 2-dichloroethane. This solvent effect on the 1-pentanol clustering is explained by the relativity of the 1-pentanol–1-pentanol interaction energy to the 1-pentanol–solvent interaction energy. Thermodynamic analysis on this solvent-induced clustering also supports that the clustering controlled by the relativity of interaction energy is inherent in the liquid state.

Web-usage mining has become the subject of intensive research, as its potential for personalized services, adaptive Web sites and customer profiling is recognized. However, the reliability of Web-usage mining results depends heavily on the proper preparation of the input datasets. In particular, errors in the reconstruction of sessions and incomplete tracing of users’ activities in a site can easily result in invalid patterns and wrong conclusions. In this study, we evaluate the performance of heuristics employed to reconstruct sessions from the server log data. Such heuristics are called to partition activities first by user and then by visit of the user in the site, where user identification mechanisms, such as cookies, may or may not be available. We propose a set of performance measures that are sensitive to two types of reconstruction errors and appropriate for different applications in knowledge discovery (KDD) applications. We have tested our framework on the Web server data of a frame-based Web site. The first experiment concerned a specific KDD application and has shown the sensitivity of the heuristics to particularities of the site's structure and traffic. The second experiment is not bound to a specific application but rather compares the performance of the heuristics for different measures and thus for different application types. Our results show that there is no single best heuristic, but our measures help the analyst in the selection of the heuristic best suited for the application at hand.

The analysis of user behavior on the Web presupposes a reliable reconstruction of the users' navigational activities. Cookies and server-generated session identifiers have been designed to allow an accurate session reconstruction. However, in the absence of reliable methods, analysts must employ heuristics (a) to identify unique visitors to a site, and (b) to distinguish among the activities of such users during independent sessions. The characteristics of the site, such as the site structure, as well as the methods used for data collection (e.g., the existence of cookies and reliable synchronization across multiple servers) may necessitate the use of different types of heuristics. In this study, we extend our work on the reliability of sessionizing mechanisms, by investigating the impact of site structure on the quality of constructed sessions. Specifically, we juxtapose sessionizing on;a frame-based and a frame-free version of a site. We investigate the behavior of cookies, server-generated session identification, and heuristics that exploit session duration, page stay time and page linkage. Different measures of session reconstruction quality, as well as experiments on the impact on the prediction of frequent entry and exit pages, show that different reconstruction heuristics can be recommended depending on the characteristics of the site. We also present first results on the impact of session reconstruction heuristics on predictive applications such as Web personalization.

A number of studies have suggested the use of discovered Web usage patterns such as asso- ciation rules, general sequential patterns, and frequent navigational paths for generating rec- ommendations in personalization systems. To- date, however, no studies have considered the conditions under which recommendation mod- els based on sequential patterns may be more appropriate for personalization as compared to those based on non-sequential patterns such as frequent itemsets. We conjecture that the structural characteristics of Web sites, such as the site topology and the degree of connectiv- ity, have a significant impact on the relative performance of these recommendation models. We present a framework for Web personaliza- tion based on association rules, contiguous and non-contiguous sequential patterns discovered from Web usage data. We then conduct a detailed comparative evaluation based on real Web usage data from three sites with different structural characteristics. Our results suggest that less constrained patterns, such as frequent itemsets, are better suited for personalization in sites with a higher degree of connectivity and shorter navigational depth, while the se- quential recommendation models may be more suitable in sites with deeper navigational depth or in sites relying on many dynamically gener- ated pages.

Web usage mining has been used effectively as an underlying mechanism for Web personalization and recommender systems. A variety of recommendation frameworks have been proposed, including some based on non-sequential models, such as association rules and clusters, and some based on sequential models, such as sequential or navigational patterns. Our recent stud- ies have suggested that the structural characteristics of Web sites, such as the site topology and the degree of connectivity, have a significant impact on the rela- tive performance of recommendation models based on association rules, contiguous and non-contiguous se- quential patterns. In this paper, we present a frame- work for a hybrid Web personalization system that can intelligently switch among different recommendation models, based on the degree of connectivity and the current location of the user within the site. We have conducted a detailed evaluation based on real Web us- age data from three sites with different structural char- acteristics. Our results show that the hybrid system selects less constrained models such as frequent item- sets when the user is navigating portions of the site with a higher degree of connectivity, while sequential recommendation models are chosen for deeper naviga- tional depths and lower degrees of connectivity. The comparative evaluation also indicates that the overall performance of hybrid system in terms of precision and coverage is better than the recommendation systems based on any of the individual models.

We describe an efficient framework for Web personalization based on sequential and non-sequential pattern discovery from usage data. Our experimental results performed on real usage data indicate that more restrictive patterns, such as contiguous sequential patterns (e.g., frequent navigational paths) are more suitable for predictive tasks, such as Web prefetching, (which involve predicting which item is accessed next by a user), while less constrained patterns, such as frequent item sets or general sequential patterns are more effective alternatives in the context of Web personalization and recommender systems.

Web usage mining, possibly used in conjunction with standard approaches to personalization such as collaborative filtering, can help address some of the shortcomings of these techniques, including reliance on subjective user ratings, lack of scalability, and poor performance in the face of high-dimensional and sparse data. However, the discovery of patterns from usage data by itself is not sufficient for performing the personalization tasks. The critical step is the effective derivation of good quality and useful (i.e., actionable) “aggregate usage profiles” from these patterns. In this paper we present and experimentally evaluate two techniques, based on clustering of user transactions and clustering of pageviews, in order to discover overlapping aggregate profiles that can be effectively used by recommender systems for real-time Web personalization. We evaluate these techniques both in terms of the quality of the individual profiles generated, as well as in the context of providing recommendations as an integrated part of a personalization engine. In particular, our results indicate that using the generated aggregate profiles, we can achieve effective personalization at early stages of users' visits to a site, based only on anonymous clickstream data and without the benefit of explicit input by these users or deeper knowledge about them.

To engage visitors to a Web site at a very early stage (i.e., before registration or authentication), personalization tools must rely primarily on clickstream data captured in Web server logs. The lack of explicit user ratings as well as the sparse nature and the large volume of data in such a setting poses serious challenges to standard collaborative filtering techniques in terms of scalability and performance. Web usage mining techniques such as clustering that rely on offline pattern discovery from user transactions can be used to improve the scalability of collaborative filtering, however, this is often at the cost of reduced recommendation accuracy. In this paper we propose effective and scalable techniques for Web personalization based on association rule discovery from usage data. Through detailed experimental evaluation on real usage data, we show that the proposed methodology can achieve better recommendation effectiveness, while maintaining a computational advantage over direct approaches to collaborative filtering such as the k -nearest-neighbor strategy.

Web usage mining, possibly used in conjunction with standard approaches to personalization such as collaborative filtering, can help address some of the shortcomings of these techniques, including reliance on subjective user ratings, lack of scalability, and poor performance in the face high- dimensional and sparse data. However, the discovery of patterns from usage data by itself is not sufficient for performing the personalization tasks. The critical step is the effective derivation of good quality and useful (i.e., actionable) "aggregate usage profiles" from these patterns. In this paper we present and experimentally evaluate two techniques, based on clustering of user transactions and clustering of pageviews, in order to discover overlapping aggregate profiles that can be effectively used by recommender systems for real-time personalization. We evaluate these techniques both in terms of the quality of the individual profiles generated, as well as in the context of providing recommendations as an integrated part of a personalization engine.