|
The wine.com API does a lot of work for you, here. I don't really see what you're accomplishing by adding a social layer on top of it. If the killer feature is the recommendation engine social or otherwise, I don't think that you're selling it well enough. I also don't think that you have a good enough feature set for a decent learning model to consistently offer up good wines. I once attempted to play around with this kind of domain space in the past. I was really disappointed at how hard it was to get the actual descriptions and star ratings of the wines I was looking at. I think the descriptions would actually be what would be useful for discovering new, awesome wines. For example, "You consistently like wines that were described in terms like honey or berry, so we suggested this Chenin Blanc to you," or, "This 2002 Cabernet Franc was suggested because you consistently favorite aged reds that are described with terms like 'tobacco' and 'leather'." If you have that, then you've created an AI wine sommelier. Start pairing things with food, and you have an app that might go places. But the data you need to make that happen is really hard to come by. Wine is also a messy topic because of the ridiculous amount of local producers, which means that in places where people actually enjoy wine, there's a good chance they're drinking something that isn't available on the API, at all. In fact, if you're a wine connoisseur, odds are you're trying to find progressively more obscure wines in the first place. In other words, why would I, as your target audience, want recommendations from anything unexhaustive? This project shows a lot of skill integrating with APIs, though, and the design isn't bad. But it seems like more of a weekend hack project than a viable service. Still, from someone who once used a beer API to teach people about search engines, well done. |
Thanks for the feedback.