Computing n-Gram Statistics in MapReduce
This work addresses the need for scalable n-gram computation in information retrieval and NLP, offering incremental improvements over existing methods.
The paper tackles the problem of efficiently computing n-gram statistics in distributed systems using MapReduce, presenting algorithms like Suffix-σ and evaluating them on datasets such as The New York Times Annotated Corpus and ClueWeb09 to expose trade-offs in performance.
Statistics about n-grams (i.e., sequences of contiguous words or other tokens in text documents or other string data) are an important building block in information retrieval and natural language processing. In this work, we study how n-gram statistics, optionally restricted by a maximum n-gram length and minimum collection frequency, can be computed efficiently harnessing MapReduce for distributed data processing. We describe different algorithms, ranging from an extension of word counting, via methods based on the Apriori principle, to a novel method Suffix-σthat relies on sorting and aggregating suffixes. We examine possible extensions of our method to support the notions of maximality/closedness and to perform aggregations beyond occurrence counting. Assuming Hadoop as a concrete MapReduce implementation, we provide insights on an efficient implementation of the methods. Extensive experiments on The New York Times Annotated Corpus and ClueWeb09 expose the relative benefits and trade-offs of the methods.