In sequence modeling, we often wish to represent complex interaction between labels, such as when performing multiple, cascaded labeling tasks on the same sequence, or when longrange dependencies exist. We present dynamic conditional random fields (DCRFs), a generalization of linear-chain conditional random fields (CRFs) in which each time slice contains a set of state variables and edges---a distributed state representation as in dynamic Bayesian networks (DBNs)---and parameters are tied across slices. Since exact

In information extraction, we often wish to identify all mentions of an entity, such as a person or organization. Traditionally, a group of words is labeled as an entity based only on local information. But information from throughout a document can be useful; for example, if the same word is used multiple times, it is likely to have the same label each time. We present a CRF that explicitly represents dependencies between the labels of pairs of similar words in a document. On a standard information extraction data set, we show that learning these dependencies leads to a 13.7% reduction in error on the field that had caused the most repetition errors. 1

The Internet presents a huge amount of useful information which is usually formatted for its users, which makes it difficult to extract relevant data from various sources. Therefore, the availability of robust, flexible Information Extraction (IE) systems that transform the Web pages into program-friendly structures such as a relational database will become a great necessity. Although many approaches for data extraction from Web pages have been developed, there has been limited effort to compare such tools. Unfortunately, in only a few cases can the results generated by distinct tools be directly compared since the addressed extraction tasks are different. This paper surveys the major Web data extraction approaches and compares them in three dimensions: the task domain, the automation degree, and the techniques used. The criteria of the first dimension explain why an IE system fails to handle some Web sites of particular structures. The criteria of the second dimension classify IE systems based on the techniques used. The criteria of the third dimension measure the degree of automation for IE systems. We believe these criteria provide qualitatively measures to evaluate various IE approaches.

Abstract Wikis are established means for the collaborative authoring, versioning and publishing of textual articles. The Wikipedia project, for example, succeeded in creating the by far largest encyclopedia just on the basis of a wiki. Recently, several approaches have been proposed on how to extend wikis to allow the creation of structured and semantically enriched content. However, the means for creating semantically enriched structured content are already available and are, although unconsciously, even used by Wikipedia authors. In this article, we present a method for revealing this structured content by extracting information from template instances. We suggest ways to efficiently query the vast amount of extracted information (e.g. more than 8 million RDF statements for the English Wikipedia version alone), leading to astonishing query answering possibilities (such as for the title question). We analyze the quality of the extracted content, and propose strategies for quality improvements with just minor modifications of the wiki systems being currently used. 1

Kernel conditional random fields (KCRFs) are introduced as a framework for discriminative modeling of graph-structured data. A representer theorem for conditional graphical models is given which shows how kernel conditional random fields arise from risk minimization procedures defined using Mercer kernels on labeled graphs. A procedure for greedily selecting cliques in the dual representation is then proposed, which allows sparse representations. By incorporating kernels and implicit feature spaces into conditional graphical models, the framework enables semi-supervised learning algorithms for structured data through the use of graph kernels.

Coreference analysis, also known as record linkage or identity uncertainty, is a difficult and important problem in natural language processing, databases, citation matching and many other tasks. This paper introduces several discriminative, conditionalprobability models for coreference analysis, all examples of undirected graphical models. Unlike many historical approaches to coreference, the models presented here are relational---they do not assume that pairwise coreference decisions should be made independently from each other. Unlike other relational models of coreference that are generative, the conditional model here can incorporate a great variety of features of the input without having to be concerned about their dependencies--- paralleling the advantages of conditional random fields over hidden Markov models. We present experiments on proper noun coreference in two text data sets, showing results in which we reduce error by nearly 28% or more over traditional thresholded record-linkage, and by up to 33% over an alternative coreference technique previously used in natural language processing.

Abstract not found

An ad hoc data source is any semistructured data source for which useful data analysis and transformation tools are not readily available. Such data must be queried, transformed and displayed by systems administrators, computational biologists, financial analysts and hosts of others on a regular basis. In this paper, we demonstrate that it is possible to generate a suite of useful data processing tools, including a semi-structured query engine, several format converters, a statistical analyzer and data visualization routines directly from the ad hoc data itself, without any human intervention. The key technical contribution of the work is a multi-phase algorithm that automatically infers the structure of an ad hoc data source and produces a format specification in the PADS data description language. Programmers wishing to implement custom data analysis tools can use such descriptions to generate printing and parsing libraries for the data. Alternatively, our software infrastructure will push these descriptions through the PADS compiler and automatically generate fully functional tools. We evaluate the performance of our inference algorithm, showing it scales linearly in the size of the training data — completing in seconds, as opposed to the hours or days it takes to write a description by hand. We also evaluate the correctness of the algorithm, demonstrating that generating accurate descriptions often requires less than 5 % of the available data. 1.

The popularity of the Web and the large number of documents available in electronic form has motivated the search for hidden knowledge in text collections. Consequently, there is growing research interest in the general topic of text mining. In this paper, we develop a text-mining system by integrating methods from Information Extraction (IE) and Data Mining (Knowledge Discovery from Databases or KDD). By utilizing existing IE and KDD techniques, text-mining systems can be developed relatively rapidly and evaluated on existing text corpora for testing IE systems. We present a general text-mining framework called DiscoTEX which employs an IE module for transforming natural-language documents into structured data and a KDD module for discovering prediction rules from the extracted data. When discovering patterns in extracted text, strict matching of strings is inadequate because textual database entries generally exhibit variations due to typographical errors, misspellings, abbreviations, and other

In this paper, we propose a machine learning approach to title extraction from general documents. By general documents, we mean documents that can belong to any one of a number of specific genres, including presentations, book chapters, technical papers, brochures, reports, and letters. Previously, methods have been proposed mainly for title extraction from research papers. It has not been clear whether it could be possible to conduct automatic title extraction from general documents. As a case study, we consider extraction from Office including Word and PowerPoint. In our approach, we annotate titles in sample documents (for Word and PowerPoint respectively) and take them as training data, train machine learning models, and perform title extraction using the trained models. Our method is unique in that we mainly utilize formatting information such as font size as features in the models. It turns out that the use of formatting information can lead to quite accurate extraction from general documents. Precision and recall for title extraction from Word is 0.810 and 0.837 respectively, and precision and recall for title extraction from PowerPoint is 0.875 and 0.895 respectively in an experiment on intranet data. Other important new findings in this work include that we can train models in one domain and apply them to another domain, and more surprisingly we can even train models in one language and apply them to another language. Moreover, we can significantly improve search ranking results in document retrieval by using the extracted titles.