The modality of textual data has been somewhat under-represented in big data and data science research thus far. This is despite the fact that large amounts of data are stored in unstructured textual format. We intend that this workshop will address this shortcoming and bring together academic and industrial researchers to exchange cutting edge research in the emerging area of extremely large-scale natural language processing (NLP). This topic has emerged in several areas in parallel in recent years: information retrieval and search engines, text mining, machine learning, web-derived corpus/computational linguistics, digital libraries, high performance and parallel computing. Common to all these areas is some or all of the main parts of the NLP pipeline: collection, cleaning, annotation, indexing, storage, retrieval and analysis of voluminous quantities of naturally occurring language data from the web or large-scale national and international digitisation initiatives. By hosting this event at IEEE Big Data 2016, we hope to encourage the communities to come together to consider synergies between NLP and data science.

In this context, numerous issues should be considered including those linked to the five Vs of big data: (a) Volume: is having more data for training and testing NLP techniques always better? (b) Variety: are all types of data available on a sufficiently large scale? (c) Velocity: how are parallel methods best applied to carry out NLP on a large scale? (d) Variability: how does inconsistent data impact on the accuracy of NLP techniques? (e) Veracity: how does the accuracy of data affect inferences that can be drawn from it?