Proposal Description

Web data has a significant potential to augment official statistical production, due to the plethora of information available on the internet. This data source has been widely explored by national statistical institutes (NSIs) in search of more timely and more relevant statistical information in response to emerging data users’ needs. In order to tap into web data, numerous endeavors have been undertaken, not only at the level of individual NSIs, but also internationally. At the European level, ESSnet Big Data I and II projects, followed by ESSnet Web Intelligence Network (WIN), gathered a fair number of NSIs who jointly with Eurostat investigated the feasibility to generate web data-based statistics within the centralized production environment.
These experiences have, on one hand, further demonstrated the potential of the use of web data in official statistics. Addressing new thematic areas, e.g., sentiment analysis in social media; acquiring missing information, e.g. in the domain of price or tourism statistics; enhancing statistical surveys and business registers, are just a few examples where web data have turned out particularly useful. On the other hand, numerous methodological and quality-related challenges, in terms of translating web data into official statistical outputs, have been strongly pronounced.
This session presents a number of facets of integrating web data into statistical production, in the context of methodological and quality standards of official statistics. It touches upon the challenges related to the data source itself, such as alternative paradigms to web data acquisitions, data source and data quality criteria and measures, methods of data classification and quality of classification models, relevance of classification frameworks. Concurrently, it discusses the assumptions underlying the attempt to develop a one-fits-all solution to produce statistics at an international level.
The above-stated topics will be illustrated with specific use-cases in different fields. The issues raised in the presentation, will be further discussed and recapitulated in the panel, moderated by the discussant, with the input from the audience.