MoodSights proprietary technology combines several major expertises to leverage consumer search data and turn these into actionable and predictive insights
This first set of expertise includes the following items: Natural Language Processing (NLP), Semantic Analysis and Multi-language capability.
- Natural Language Processing (NLP) is used for tasks such as heading detection, language detection, key phrase extraction and classification of words and word combinations.
- Semantic analysis. Semantic analysis is major in the context of query analysis, because many homonyms or errors are used by consumers. For example, fine and fine, London / Londom, holidays / holydays.
- All of these technologies are currently usable for all Indo-European languages and can eventually be adapted to Arabic and Asian languages
This second set brings together the following know-how: Analysis of the understanding of interests / concerns, Mass production and high-speed analysis, Geolocation grid.
- Al / Algorithms. Automated and deployed on a large scale, these technologies allow MoodSights to process huge amounts of information, to update them regularly, within short deadline, and at very attractive production costs compared to more traditional methods.
- Based on data from search engine datasets, MoodSights geolocates the keywords and word combinations used by consumers at a varying degree (as focused as a town or enlarged as a country)
This last brick brings together all the quantitative expertise of MoodSights: Calculation of metadata and construction of trends, Back-tests and consistency checks, construction of Scores.
- Meta data. MoodSights is able to differentiate general searches on a brand (I’m looking for basic information on ESPN, like the evening program for example) vs. more specific research which indicates that a consumer / internet user is in the pre-purchase / pre-subscription phase (I am looking for information on prices, types of ESPN offers, I compare with competitors, etc.)
- In the case of a general brand attractiveness score, the 2 types of search data will be taken into account. In the case of an intent of purchase score, only the second type of search data will be retained