3 ms·

This kid would definitely make a great marketing person. There's absolutely nothing new in what he's done but he's presenting it really well, thumbs up to his

This kid would definitely make a great marketing person. There's absolutely nothing new in what he's done but he's presenting it really well, thumbs up to his parents who have probably done quite a bit of the work ;) When I was 18 (10 years ago) I did loads of research into that stuff and knew just as much about information retrieval, vector space tf-ids models, latent semantic indexing, wordnet analysis, etc. At the time it was fairly cutting edge research. This stuff isn't anything new now, I was actually forced to decipher some research papers instead of reading popular books on the subject. It was fairly obvious to me back then that none of these techniques worked well for general web search. I did end up building a system that clustered google search results (in realtime) into DMOZ categories letting you refine your search results by clicking a category (which was actually useful and worked quite well in case you were searching for something ambiguous like "jaguar"). None of these techniques are new to anyone working in information retrieval. Just looking at co-occurrence of words in tweets and expanding the query with some related terms (weighted appropriately) would probably achieve what he has done (weekend project for an average dev). I'd call this kid really smart if he'd actually figure out how to improve general web search, or could think of a useful application at least. Talking about existing research and making it look like your own isn't great form in my opinion. Coming up with your own definition for a "word" just makes you look stuck up. Much better off acknowledging work of other researchers and quoting them, although that would never generate as much press I guess. Sorry if the rant is quite negative, I'm just getting a bit fed up with all the marketing surrounding "young geniuses" and teenage entrepreneurs these days. If I wanted to read that stuff I'd get the local paper.

nschiefer 14y ago
Most of your feedback is perfectly fair. My algorithm was not meant for general search, not am I a "young genius" or anything of the sort. Similarly, techniques like tf-idf, latent semantic analysis, document modelling, and even pseudo-relevance feedback expansion are no longer "cutting edge" techniques. However, your blanket characterization that "there's absolutely nothing new" in my work and that I just talked "about existing research" while "making it look like [my] own" is somewhat offensive. Based on a fairly extensive review of the literature, the algorithm that I developed is novel and seems to outperform a number of these standard techniques on short document like tweets. As for "just looking at co-occurrence," that's essentially a type of pseudo-relevance feedback expansion and is, of course, well known and easy to implement. Please realize the difference between a news interview published online and the paper I submitted to the fair when assessing the novelty of my work and when suggesting that I made no reference to others' work. Regarding my "definition for a 'word'", I apologize for appearing pretentious. I was asked to speak on the conference's theme of "redefinition" in relation to my research and did the best I could. Finally, I think it's kind of strange that you automatically assumed that my parents even worked on my project. Neither of them are even familiar with the details of the project. My work was my own, and your automatic assumption of what is tantamount of broad academic fraud and plagiarism assumed particularly bad faith. In any case, thanks for your honest feedback; I try to be careful about how I come across and I'll be even more mindful in the future.

nschiefer 14y ago

Most of your feedback is perfectly fair. My algorithm was not meant for general search, not am I a "young genius" or anything of the sort. Similarly, techniques like tf-idf, latent semantic analysis, document modelling, and even pseudo-relevance feedback expansion are no longer "cutting edge" techniques. However, your blanket characterization that "there's absolutely nothing new" in my work and that I just talked "about existing research" while "making it look like [my] own" is somewhat offensive. Based on a fairly extensive review of the literature, the algorithm that I developed is novel and seems to outperform a number of these standard techniques on short document like tweets. As for "just looking at co-occurrence," that's essentially a type of pseudo-relevance feedback expansion and is, of course, well known and easy to implement. Please realize the difference between a news interview published online and the paper I submitted to the fair when assessing the novelty of my work and when suggesting that I made no reference to others' work. Regarding my "definition for a 'word'", I apologize for appearing pretentious. I was asked to speak on the conference's theme of "redefinition" in relation to my research and did the best I could. Finally, I think it's kind of strange that you automatically assumed that my parents even worked on my project. Neither of them are even familiar with the details of the project. My work was my own, and your automatic assumption of what is tantamount of broad academic fraud and plagiarism assumed particularly bad faith. In any case, thanks for your honest feedback; I try to be careful about how I come across and I'll be even more mindful in the future.