Web spam can significantly deteriorate the quality of search engine results. Thus there is a large incentive for commercial search engines to detect spam pages efficiently and accurately. In this paper we present a spam detection system that uses the topology of the Web graph by exploiting the link dependencies among the Web pages, and the content of the pages themselves. We find that linked hosts tend to belong to the same class: either both are spam or both are non-spam. We demonstrate three methods of incorporating the Web graph topology into the predictions obtained by our base classifier: (i) clustering the host graph, and assigning the label of all hosts in the cluster by majority vote, (ii) propagating the predicted labels to neighboring hosts, and (iii) using the predicted labels of neighboring hosts as new features and retraining the classifier. The result is an accurate system for detecting Web spam that can be applied in practice to large-scale Web data.

We perform a statistical analysis of a large collection of Web pages, focusing on spam detection. We study several metrics such as degree correlations, number of neighbors, rank propagation through links, TrustRank and others to build several automatic web spam classifiers. This paper presents a study of the performance of each of these classifiers alone, as well as their combined performance. Using this approach we are able to detect 80.4% of the Web spam in our sample, with only 1.1% of false positives.

We describe the WEBSPAM-UK2006 collection, a large set of Web pages that have been manually annotated with labels indicating if the hosts are include Web spam aspects or not. This is the first publicly available Web spam collection that includes page contents and links, and that has been labelled by a large and diverse set of judges. 1

This paper introduces a family of link-based ranking algorithms that propagate page importance through links. In these algorithms there is a damping function that decreases with distance, so a direct link implies more endorsement than a link through a long path. PageRank is the most widely known ranking function of this family. The main objective of this paper is to determine whether this family of ranking techniques has some interest per se, and how different choices for the damping function impact on rank quality and on convergence speed. Even though our results suggest that Page-Rank can be approximated with other simpler forms of rankings that may be computed more efficiently, our focus is of more speculative nature, in that it aims at separating the kernel of PageRank, that is, link-based importance propagation, from the way propagation decays over paths. We focus on three damping functions, having linear, exponential, and hyperbolic decay on the lengths of the paths. The exponential decay corresponds to PageRank, and the other functions are new. Our presentation includes algorithms, analysis, comparisons and experiments that study their behavior under different parameters in real Web graph data. Among other results, we show how to calculate a linear approximation that induces a page ordering that is almost identical to Page-Rank’s using a fixed small number of iterations; comparisons were performed using Kendall’s τ on large domain datasets.

We propose link-based techniques for automating the detection of Web spam, a term referring to pages which use deceptive techniques to obtain undeservedly high scores in search engines. The issue of Web spam is widespread and difficult to solve, mostly due to the large size of the Web which means that, in practice, many algorithms are infeasible. We perform a statistical analysis of a large collection of Web pages. In particular, we compute statistics of the links in the vicinity of every Web page applying rank propagation and probabilistic counting over the entire Web graph in a scalable way. We build several automatic web spam classifiers using different techniques. This paper presents a study of the performance of each of these classifiers alone, as well as their combined performance. Based on these results we propose spam detection techniques which only consider the link structure of Web, regardless of page contents. These statistical features are used to build a classifier that is tested over a large collection of Web link spam. After ten-fold cross-validation, our best classifiers have a performance comparable to that of state-of-the-art spam classifiers that use content attributes, and orthogonal to their methods.

So fast, so cheap, so efficient, Internet is nowadays incontestably communication mean of choice for personal, business and academic purposes. Unfortunately, Internet has not only this beautiful face. Malicious activities enjoy as well this so fast, cheap and efficient mean. The last decade, Internet worms took the lights. In the recent years, spams are invading one of the most used services of Internet: email. This paper summarizes most of techniques used to filter spams by analyzing the email content. Povzetek: Članek pregledno opisuje metode za filtriranje elektronske pošte. 1

This extended abstract describes and analyses a near-optimal probabilistic algorithm, HYPERLOGLOG, dedicated to estimating the number of distinct elements (the cardinality) of very large data ensembles. Using an auxiliary memory of m units (typically, “short bytes”), HYPERLOGLOG performs a single pass over the data and produces an estimate of the cardinality such that the relative accuracy (the standard error) is typically about 1.04 / √ m. This improves on the best previously known cardinality estimator, LOGLOG, whose accuracy can be matched by consuming only 64% of the original memory. For instance, the new algorithm makes it possible to estimate cardinalities well beyond 10 9 with a typical accuracy of 2 % while using a memory of only 1.5 kilobytes. The algorithm parallelizes optimally and adapts to the sliding window model.

In this paper, we study the classification of web spam. Web spam refers to pages that use techniques to mislead search engines into assigning them higher rank, thus increasing their site traffic. Our contributions are two fold. First, we find that the method of dataset construction is crucial for accurate spam classification and we note that this problem occurs generally in learning problems and can be hard to detect. In particular, we find that ensuring no overlapping domains between test and training sets is necessary to accurately test a web spam classifier. In our case, classification performance can differ by as much as 40 % in precision when using non-domain-separated data. Second, we show ranktime features can improve the performance of a web spam classifier. Our paper is the first to investigate the use of rank-time features, and in particular query-dependent ranktime features, for web spam detection. We show that the use of rank-time and query-dependent features can lead to an increase in accuracy over a classifier trained using page-based content only.

Link spam refers to attempts to promote the ranking of spammers ’ web sites by deceiving link-based ranking algorithms in search engines. Spammers often create densely connected link structure of sites so called “link farm”. In this paper, we study the overall structure and distribution of link farms in a large-scale graph of the Japanese Web with 5.8 million sites and 283 million links. To examine the spam structure, we apply three graph algorithms to the web graph. First, the web graph is decomposed into strongly connected components (SCC). Beside the largest SCC (core) in the center of the web, we have observed that most of large components consist of link farms. Next, to extract spam sites in the core, we enumerate maximal cliques as seeds of link farms. Finally, we expand these link farms as a reliable spam seed set by a minimum cut technique that separates links among spam and non-spam sites. We found about 0.6 million spam sites in SCCs around the core, and extracted additional 8 thousand and 49 thousand sites as spams with high precision in the core by the maximal clique enumeration and by the minimum cut technique, respectively. 1.

Assessing the trustworthiness of reviews is a key issue for the maintainers of opinion sites such as TripAdvisor. In this paper we propose a distortion criterion for assessing the impact of methods for uncovering suspicious hotel reviews in TripAdvisor. The principle is that dishonest reviews will distort the overall popularity ranking for a collection of hotels. Thus a mechanism that deletes dishonest reviews will distort the popularity ranking significantly, when compared with the removal of a similar set of reviews at random. This distortion can be quantified by comparing popularity rankings before and after deletion, using rank correlation. We present an evaluation of this strategy in the assessment of shill detection mechanisms on a dataset of hotel reviews collected from TripAdvisor.