Connect the Last Mile - Textual Entailment Identify Efforts in Bridging the Digital Divide

Zhuangyuan Fan,Fan Zhang
DOI: https://doi.org/10.1007/s44212-022-00010-4
2022-01-01
Urban Informatics
Abstract:With many people’s life and work more and more Internet-dependent, we yet have to face the persistent digital divide. In the U.S., funding programs in the past allocated resources to incumbent Internet Service Providers (ISP) who failed to connect the rural and low-income communities. This is largely due to an information mismatch - the financially strong ISPs do not necessarily have the vision to connect the marginalized community. This project sets out to build a multidimensional profile database of ISPs that emphasizes the potential of serving the underserved or unserved areas. By applying the Textual Entailment Natural Language Processing algorithm, we infer the vision, goals, and identity of more than 1,000 ISPs using a large number of self-description texts from their websites. Then we created an ISP typology based on the socio-economic profiles of the service areas of these ISPs and their business vision. Deviating from the traditional bifurcating definition of “for-profit” and “non-profits” business, this ISP typology illustrates evident geographical, social, and service quality level division of the current Internet services in the U.S. Planners and policymakers could use this typology to create specific funding programs that effectively address the current digital divide.
What problem does this paper attempt to address?