In textual conversations, it is often difficult for people to gauge emotions. This is particularly true when one is starting to get to know someone and are unable to decipher his or her textual messaging style. However, the implications of being able to infer the meaning or emotions of words and sentences are important, as it drives powerful use cases for people dealing with mental problems, going through difficult periods in their lives, or even for something as familiar as customer engagement.
Techsumption recently collaborated with a renowned research institute with the aim to collect conversational data in
The project involved recruiting a total of 3 distinguished experts and 60 participants to engage in a 30 minutes conversation each. Participants and experts are both anonymized and carried out conversations revolving around specific topics pre-defined by our client.
The crux and difficulty of this project lie in identifying the right experts and respondents and ensuring a fruitful conversational exchange. Through multiple project meetings, the best profiles were selected and an entire workflow for expert and respondent recruitment, data collection and annotation was surfaced to
At each phase of the project delivery, care was taken to verify and validate preliminary findings, ensuring no surprises when the final datasets are furnished.
The project was successfully concluded with the intended outcome, paving the way for the client to use the data collected and annotated in the training of machines/programs which are able to infer and decipher sentences and words from a given textual conversation.