This is the home of the SMILE training and testing coprus of Impact Models.
Motivation: The social services sector, rapidly growing and difficult to understand, spans across multiple languages, with different vocabularies and varying degrees of detail. A key example facing the social services sector in resource planning is determining the social services that are readily available, particularly, identifying the service providers in a given geographic area, the services they provide, and whom they are provided. A second challenge is identifying the best social purpose organizations to serve a client based on their characteristics and needs.
To address these shortcomings, information about social purpose organizations (SPOs) and the funds they receive must be identified, including the programs and services these organizations provide, targeted client needs, client characteristics, expected outcomes, and the distribution of funding. These details are collectively termed as an “Impact Model”. SeMantIc roLe Extraction (SMILE) is an architecture that processes unstructured text and extract actionable Impact Models for social services. It will use and extend an ontology for the social services domain to extract key Impact Models concepts and their relationships.
Goal: Develop a Natural Language Understanding (NLU) system that can find and extract, from web documents, artifacts of the Impact Model, as defined by the Common Impact Data Standard (CIDS) and developed at the CSSE. CSSE aims to enhance social services’ effectiveness and efficiency. CIDS has recently been announced as the Employment and Social Development Canada’s data standard for defining and reporting on impact models for over 800 social service funders across Canada. The proposed research will extract impact model information from unstructured textual data scattered across sources.
The objective of the SMILE Annotation project are: Create the Impact Model training dataset. Create a database of Social Purpose Organizations across Canada.