Abstract-Automatic semantic annotation of data from databases or the web is an important pre-process for data cleansing and record linkage. It can be used to resolve the problem of imperfect field alignment in a database or identify comparable fields for matching records from multiple sources. The annotation process is not trivial because data values may be noisy, such as abbreviations, variations or misspellings. In particular, overlapping features usually exist in a lexicon-based approach. In this work, we present a probabilistic address parser based on linear-chain conditional random fields (CRFs), which allow more expressive token-level features compared to hidden Markov models (HMMs). In additions, we also proposed two general enhancement techniques to improve the performance. One is taking original semi-structure of the data into account. Another is post-processing of the output sequences of the parser by combining its conditional probability and a score function, which is based on a learned stochastic regular grammar (SRG) that captures segment-level dependencies. Experiments were conducted by comparing the CRF parser to a HMM parser and a semi-Markov CRF parser in two real-world datasets. The CRF parser out-performed the HMM parser and the semiMarkov CRF in both datasets in terms of classification accuracy. Leveraging the structure of the data and combining the linearchain CRF with the SRG further improved the parser to achieve an accuracy of 97% on a postal dataset and 96% on a company dataset.

Abstract. Probabilistic record linkage is a well established topic in the literature. Fellegi-Sunter probabilistic record linkage and its enhanced versions are commonly used methods, which calculate match and nonmatch weights for each pair of records. Bayesian network classifiersnaive Bayes classifier and TAN have also been successfully used here. Recently, an extended version of TAN (called ETAN) has been developed and proved superior in classification accuracy to conventional TAN. However, no previous work has applied ETAN to record linkage and investigated the benefits of using naturally existing hierarchical feature level information and parsed fields of the datasets. In this work, we extend the naive Bayes classifier with such hierarchical feature level information. Finally we illustrate the benefits of our method over previously proposed methods on 4 datasets in terms of the linkage performance (F1 score). We also show the results can be further improved by evaluating the benefit provided by additionally parsing the fields of these datasets.

No abstract