AT&T Labs

About: AT&T Labs is a based out in . It is known for research contribution in the topics: Network packet & The Internet. The organization has 1879 authors who have published 5595 publications receiving 483151 citations.

Hubbub: a sound-enhanced mobile instant messenger that supports awareness and opportunistic interactions

Abstract: There have been many attempts to support awareness and lightweight interactions using video and audio, but few have been built on widely available infrastructure. Text-based systems have become more popular, but few support awareness, opportunistic conversations, and mobility, three important elements of distributed collaboration. We built on the popularity of text-based Instant Messengers (IM) by building a mobile IM called Hubbub that tries to provide all three, notably through the use of earcons. In a 5.5-month use study, we found that Hubbub helped people feel connected to colleagues in other locations and supported opportunistic interactions. The sounds provided effective awareness cues, although some found them annoying. It was more important to support graceful transitions between multiple fixed locations than to support wireless access, although both were useful

Exploiting a probabilistic hierarchical model for generation

Abstract: Previous stochastic approaches to generation do not include a tree-based representation of syntax. While this may be adequate or even advantageous for some applications, other applications profit from using as much syntactic knowledge as is available, leaving to a stochastic model only those issues that are not determined by the grammar. We present initial results showing that a tree-based model derived from a tree-annotated corpus improves on a tree model derived from an unannotated corpus, and that a tree-based stochastic model with a hand-crafted grammar outperforms both.

Experience in measuring backbone traffic variability: models, metrics, measurements and meaning

Abstract: Understanding the variability of Internet traffic in backbone networks is essential to better plan and manage existing networks, as well as to design next generation networks. However, most traffic analyses that might be used to approach this problem are based on detailed packet or flow level measurements, which are usually not available throughout a large network. As a result there is a poor understanding of backbone traffic variability, and its impact on network operations (e.g. on capacity planning or traffic engineering).This paper introduces a metric for measuring backbone traffic variability that is grounded on simple but powerful traffic theory. What sets this metric apart, however, is that we present a method for making practical measurements of the metric using widely available SNMP traffic measurements. Furthermore, we use a novel method to overcome the major limitation of SNMP measurements -- that they only provide link statistics. The method, based on a "gravity model", derives an approximate traffic matrix from the SNMP data. In addition to simulations, we use more than 1 year's worth of SNMP data from an operational IP network of about 1000 nodes to test our methods. We also delve into the degree and sources of variability in real backbone traffic, providing insight into the true nature of traffic variability.

Tree Pattern Relaxation

Abstract: Tree patterns are fundamental to querying tree-structured data like XML Because of the heterogeneity of XML data, it is often more appropriate to permit approximate query matching and return ranked answers, in the spirit of Information Retrieval, than to return only exact answers In this paper, we study the problem of approximate XML query matching, based on tree pattern relaxations, and devise efficient algorithms to evaluate relaxed tree patterns We consider weighted tree patterns, where exact and relaxed weights, associated with nodes and edges of the tree pattern, are used to compute the scores of query answers We are interested in the problem of finding answers whose scores are at least as large as a given threshold We design data pruning algorithms where intermediate query results are filtered dynamically during the evaluation process We develop anoptimization that exploits scores of intermediate results to improve query evaluation efficiency Finally, we show experimentally that our techniques outperform rewriting-based and post-pruning strategies

Using suffix arrays to compute term frequency and document frequency for all substrings in a corpus

Abstract: Bigrams and trigrams are commonly used in statistical natural language processing; this paper will describe techniques for working with much longer n-grams. Suffix arrays (Manber and Myers 1990) were first introduced to compute the frequency and location of a substring (n-gram) in a sequence (corpus) of length N. To compute frequencies over all N(N + 1)/2 substrings in a corpus, the substrings are grouped into a manageable number of equivalence classes. In this way, a prohibitive computation over substrings is reduced to a manageable computation over classes. This paper presents both the algorithms and the code that were used to compute term frequency (tf) and document frequency (dr)for all n-grams in two large corpora, an English corpus of 50 million words of Wall Street Journal and a Japanese corpus of 216 million characters of Mainichi Shimbun.The second half of the paper uses these frequencies to find "interesting" substrings. Lexicographers have been interested in n-grams with high mutual information (MI) where the joint term frequency is higher than what would be expected by chance, assuming that the parts of the n-gram combine independently. Residual inverse document frequency (RIDF) compares document frequency to another model of chance where terms with a particular term frequency are distributed randomly throughout the collection. MI tends to pick out phrases with noncompositional semantics (which often violate the independence assumption) whereas RIDF tends to highlight technical terminology, names, and good keywords for information retrieval (which tend to exhibit nonrandom distributions over documents). The combination of both MI and RIDF is better than either by itself in a Japanese word extraction task.