We show that the beta process is the de Finetti mixing distribution underlying the Indian buffet process of [2]. This result shows that the beta process plays the role for the Indian buffet process that the Dirichlet process plays for Chinese restaurant process, a parallel that guides us in deriving analogs for the beta process of the many known extensions of the Dirichlet process. In particular we define Bayesian hierarchies of beta processes and use the connection to the beta process to develop posterior inference algorithms for the Indian buffet process. We also present an application to document classification, exploring a relationship between the hierarchical beta process and smoothed naive Bayes models. 1 1

Multinomial distributions are often used to model text documents. However, they do not capture well the phenomenon that words in a document tend to appear in bursts: if a word appears once, it is more likely to appear again. In this paper, we propose the Dirichlet compound multinomial model (DCM) as an alternative to the multinomial. The DCM model has one additional degree of freedom, which allows it to capture burstiness. We show experimentally that the DCM is substantially better than the multinomial at modeling text data, measured by perplexity. We also show using three standard document collections that the DCM leads to better classification than the multinomial model. DCM performance is comparable to that obtained with multiple heuristic changes to the multinomial model. 1.

With the growing importance of systems containing arbitrary semistructured relationships, the need for supporting users searching in such repositories has grown. Currently support for users' search needs either has required domain-specific user interfaces or has required users to be schema experts. We have developed a generalpurpose tool that offers users helpful navigation and refinement options for seeking information in these semistructured repositories. We show how a tool can be built without requiring domain-specific assumptions about the information being explored. In addition to describing a general approach to the problem, we provide a set of natural, general-purpose refinement tactics, many generalized from past work on textual information retrieval.

In this paper we present a method for reasoning about privacy using the concepts of exchangeability and deFinetti’s theorem. We illustrate the usefulness of this technique by using it to attack a popular data sanitization scheme known as Anatomy. We stress that Anatomy is not the only sanitization scheme that is vulnerable to this attack. In fact, any scheme that uses the random worlds model, i.i.d. model, or tuple-independent model needs to be re-evaluated. The difference between the attack presented here and others that have been proposed in the past is that we do not need extensive background knowledge. An attacker only needs to know the nonsensitive attributes of one individual in the data, and can carry out this attack just by building a machine learning model over the sanitized data. The reason this attack is successful is that it exploits a subtle flaw in the way prior work computed the probability of disclosure of a sensitive attribute. We demonstrate this theoretically, empirically, and with intuitive examples. We also discuss how this generalizes to many other privacy schemes.

In a survey we found that most developers have experienced duplicated bug reports, however, only few considered them as a serious problem. This contradicts popular wisdom that considers bug duplicates as a serious problem for open source projects. In the survey, developers also pointed out that the additional information provided by duplicates helps to resolve bugs quicker. In this paper, we therefore propose to merge bug duplicates, rather than treating them separately. We quantify the amount of information that is added for developers and show that automatic triaging can be improved as well. In addition, we discuss the different reasons why users submit duplicate bug reports in the first place. 1.

The problem that this thesis concentrates on is phishing attacks. Phishing attacks use email messages and web sites designed to look as if they come from a known and legitimate organization, in order to deceive users into submitting their personal, financial, or computer account information online at those fake web sites. Phishing is a semantic attack. The fundamental problem of phishing is that when a user submits sensitive information online under an attack, his mental model about this submission is different from the system model that actually performs this submission. Specifically, the system sends the data to a different web site from the one where the user intends to submit the data. The fundamental solution to phishing is to bridge the semantic gap between the user’s mental model and the system model. The user interface is where human users interact with the computer system. It is where a user’s intention transforms into a system operation. It is where the semantic gap happens under phishing attacks. And therefore, it is where the phishing should be solved.

Many different topic models have been used successfully for a variety of applications. However, even state-of-the-art topic models suffer from the important flaw that they do not capture the tendency of words to appear in bursts; it is a fundamental property of language that if a word is used once in a document, it is more likely to be used again. We introduce a topic model that uses Dirichlet compound multinomial (DCM) distributions to model this burstiness phenomenon. On both text and non-text datasets, the new model achieves better held-out likelihood than standard latent Dirichlet allocation (LDA). It is straightforward to incorporate the DCM extension into topic models that are more complex than LDA. 1.

Informal communication (e-mail, bulletin boards) poses a difficult learning environment because traditional grammatical and lexical information are noisy. Other information is necessary for tasks such as named entity detection. How topic-centric, or informative, a word is can be valuable information. It is well known that informative words are best modeled by “heavy-tailed ” distributions, such as mixture models. However, informativeness scores do not take full advantage of this fact. We introduce a new informativeness score that directly utilizes mixture model likelihood to identify informative words. We use the task of extracting restaurant names from bulletin board posts as a way to determine effectiveness. We find that our “mixture score ” is weakly effective alone and highly effective when combined with Inverse Document Frequency. We compare against other informativeness criteria and find that only Residual IDF is competitive against our combined IDF/Mixture score.

Abstract. Compression-based text classification methods are easy to apply, requiring virtually no preprocessing of the data. Most such methods are character-based, and thus have the potential to automatically capture non-word features of a document, such as punctuation, wordstems, and features spanning more than one word. However, compressionbased classification methods have drawbacks (such as slow running time), and not all such methods are equally effective. We present the results of a number of experiments designed to evaluate the effectiveness and behavior of different compression-based text classification methods on English text. Among our experiments are some specifically designed to test whether the ability to capture non-word (including super-word) features causes character-based text compression methods to achieve more accurate classification. 1

Our emotional state influences our choices. Research on how it happens usually comes from the lab. We know relatively little about how real world emotions affect real world settings, like financial markets. Here, we demonstrate that estimating emotions from weblogs provides novel information about future stock market prices. That is, it provides information not already apparent from market data. Specifically, we estimate anxiety, worry and fear from a dataset of over 20 million posts made on the site LiveJournal. Using a Granger-causal framework, we find that increases in expressions of anxiety, evidenced by computationally-identified linguistic features, predict downward pressure on the S&P 500 index. We also present a confirmation of this result via Monte Carlo simulation. The findings show how the mood of millions in a large online community, even one that primarily discusses daily life, can anticipate changes in a seemingly unrelated system. Beyond this, the results suggest new ways to gauge public opinion and predict its impact.