An indicator of the presence of an opinion and its polarity are the words immediately surrounding a potential opinion "target". But not all the words near the target are likely to be relevant to finding an opinion.

This talk describes work in retrieving opinion words and their links to within-sentence targets in an information technology (IT) business corpus through crowdsourcing. The opinion words for which we are looking are ones that apply to specific IT business concepts relevant to a larger research project in the social diffusion of innovations. Existing resources do not fully cover this domain.

I will present a data collection pipeline and a user interface that avoids some of the pitfalls of asking untrained individuals about word-target links. Our user interface evades some of problems caused by the inherent subjectivity of the task. I will also briefly describe one of the "downstream" uses for the data in a machine learning technique to acquire syntactic features for sentiment classification.