What the World Wants determines what products people want inventors to make. We do this by analyzing public data on social media sites using natural language processing. For example, many social media users expressed the desire for an app that prevents drunk texting. We aggregate this data and group together data that express the same concept. On http://awesomeapp.tk, we display the most popular concepts.
We came up with this idea because one of the toughest part of creating a successful product is coming up with a good idea. What the World Wants helps entrepreneurs and developers discover what products users want.
The feature that we're most proud of is grouping together similar concepts. We began by using a part-of-speech tagger to filter out sentences that aren't like "I want an app that verb noun." Next, we used WordNet to replace some synonyms and a stemming algorithm to get the root of words. Then, we used the tf–idf algorithm combined with a bigram corpus to calculate the structure similarity between two sentences. After computing the similarities, we used DFS to group together concepts with a similarity above a certain threshold. Finally, we compare the grouped concepts to generate a single sentence that describes the concept. We were able to generate over 300 concepts groups with up to 51 grouped concepts each.