Many important problems involve clustering large datasets. Although naive implementations of clustering are computationally expensive, there are established efficient techniques for clustering when the dataset has either (1) a limited number of clusters, (2) a low feature dimensionality, or (3) a small number of data points. However, there has been much less work on methods of efficiently clustering datasets that are large in all three ways at once, for example, having millions of data points that exist in many thousands of dimensions representing many thousands of clusters. We present a new technique for clustering these large, high-dimensional datasets. The key idea involves using a cheap, approximate distance measure to efficiently divide the data into overlapping subsets we call canopies. Then clustering is performed by measuring exact distances only between points that occur in a common canopy. Using canopies, large clustering problems that were formerly impossible become practical. Under reasonable assumptions about the cheap distance metric, this reduction in computational cost comes without any loss in clustering accuracy. Canopies can be applied to many domains and used with a variety of clustering approaches, including Greedy Agglomerative Clustering, K-means and Expectation-Maximization. We present experimental results on grouping bibliographic citations from the reference sections of research papers. Here the canopy approach reduces computation time over a traditional clustering approach by more than an order of magnitude and decreases error in comparison to a previously used algorithm by 25%.

Deduplication is a key operation in integrating data from multiple sources. The main challenge in this task is designing a function that can resolve when a pair of records refer to the same entity in spite of various data inconsistencies. Most existing systems use hand-coded functions. One way to overcome the tedium of hand-coding is to train a classifier to distinguish between duplicates and non-duplicates. The success of this method critically hinges on being able to provide a covering and challenging set of training pairs that bring out the subtlety of the deduplication function. This is non-trivial because it requires manually searching for various data inconsistencies between any two records spread apart in large lists. We present our design of a learning-based deduplication system that uses a novel method of interactively discovering challenging training pairs using active learning. Our experiments on real-life datasets show that active learning signicantly reduces the number of instances needed to achieve high accuracy. We investigate various design issues that arise in building a system to provide interactive response, fast convergence, and interpretable output.

Detecting database records that are approximate duplicates, but not exact duplicates, is an important task. Databases may contain duplicate records concerning the same realworld entity because of data entry errors, because of unstandardized abbreviations, or because of differences in the detailed schemas of records from multiple databases, among other reasons. In this paper, we present an efficient algorithm for recognizing clusters of approximately duplicate records. Three key ideas distinguish the algorithm presented. First, a version of the Smith-Waterman algorithm for computing minimum edit-distance is used as a domainindependent method to recognize pairs of approximately duplicate records. Second, the union/find algorithm is used to keep track of clusters of duplicate records incrementally, as pairwise duplicate relationships are discovered. Third, the algorithm uses a priority queue of cluster subsets to respond adaptively to the size and homogeneity of the clusters discovered as...

When integrating information from multiple websites, the same data objects can exist in inconsistent text formats across sites, making it di#cult to identify matching objects using exact text match. We have developed an object identification system called Active Atlas, which compares the objects' shared attributes in order to identify matching objects. Certain attributes are more important for deciding if a mapping should exist between two objects. Previous methods of object identification have required manual construction of object identification rules or mapping rules for determining the mappings between objects. This manual process is time consuming and error-prone.

Data integration solutions dealing with large amounts of data have been strongly required in the last few years. Besides the traditional data integration problems (e.g. schema integration, local to global schema mappings), three additional data problems have to be dealt with: (1) the absence of universal keys across dierent databases that is known as the object identity problem, (2) the existence of keyboard errors in the data, and (3) the presence of inconsistencies in data coming from multiple sources. Dealing with these problems is globally called the data cleaning process. In this work, we propose a framework which oers the fundamental services required by this process: data transformation, duplicate elimination and multi-table matching. These services are implemented using a set of purposely designed macro-operators. Moreover, we propose an SQL extension for specifying each of the macro-operators. One important feature of the framework is the ability of explicitly includ...

In this paper we review studies of the growth of the Internet and technologies that are useful for information search and retrieval on the Web. We present data on the Internet from several different sources, e.g., current as well as projected number of users, hosts, and Web sites. Although numerical figures vary, overall trends cited

Data cleaning is a vital process that ensures the quality of data stored in real-world databases. Data cleaning problems are frequently encountered in many research areas, such as knowledge discovery in databases, data warehousing, system integration and e-services. The process of identifying the record pairs that represent the same entity (duplicate records), commonly known as record linkage, is one of the essential elements of data cleaning. In this paper, we address the record linkage problem by adopting a machine learning approach. Three models are proposed and are analyzed empirically. Since no existing model, including those proposed in this paper, has been proved to be superior, we have developed an interactive Record Linkage Toolbox named TAILOR. Users of TAILOR can build their own record linkage models by tuning system parameters and by plugging in in-house developed and public domain tools. The proposed toolbox serves as a framework for the record linkage process, and is designed in an extensible way to interface with existing and future record linkage models. We have conducted an extensive experimental study to evaluate our proposed models using not only synthetic but also real data. Results show that the proposed machine learning record linkage models outperform the existing ones both in accuracy and in performance.

Record linkage is the task of quickly and accurately identifying records corresponding to the same entity from one or more data sources. Record linkage is also known as data cleaning, entity reconciliation or identification and the merge/purge problem. This paper presents the "standard" probabilistic record linkage model and the associated algorithm. Recent work in information retrieval, federated database systems and data mining have proposed alternatives to key components of the standard algorithm. The impact of these alternatives on the standard approach are assessed. The key question is whether and how these new alternatives are better in terms of time, accuracy and degree of automation for a particular record linkage application.

Detecting database records that are approximate duplicates, but not exact duplicates, is an important task. Databases may contain duplicate records concerning the same real-world entity because of data entry errors, unstandardized abbreviations, or differences in the detailed schemas of records from multiple databases – such as what happens in data warehousing where records from multiple data sources are integrated into a single source of information – among other reasons. In this paper we review a system to detect approximate duplicate records in a database and provide properties that a pair-wise record matching algorithm must have in order to have a successful duplicate detection system. 1

The first step prior to data mining is often to merge databases from different sources. Entries in these databases or descriptions retrieved using information extraction. may use significantly different vocabularies, so one often needs to determine whether similar descriptions refer to the same item or to different items (e.g., people or goods). String edit distance is an elegant way of defining the degree of similarity between entries and can be efficiently computed using dynamic programming (Ristad and Yianilos, 1977). However, in order to achieve reasonable accuracy, most real problems require the use of extended sets of edit rules with associated costs that are tuned specifically to each data set. We present a flexible approach to string edit distance, which can be automatically tuned to different data sets and can use synonym dictionaries. Dynamic programming is used to calculate the edit distance between a pair of strings based on a set of string edit rules including a new edit rule that allows words and phrases to be deleted or substituted. A genetic algorithm is used to learn costs corresponding to each edit rule based on a small set of labeled training data. Deleting contentless words like "method " and substituting synonyms such as "ibuprofen " for "Motrin " significantly increases the algorithm’s accuracy (from 80 % to 90 % on a difficult sample medical data set), when costs are correctly tuned. This string edit-based matching tool is easily adapted for a variety of different cases when one needs to recognize which text strings from different information sources refer to the same item such as a person, address, medical procedure or product.