I recently built slangthesaurus.com, a thesaurus… for slang. It’s not perfect by any means, but it’s a very cool demo of what is possible with some basic machine learning. The thing is, it could be much better, and I have some other things I’m working on right now.

There are several things I can focus on to improve this: making the synonyms better (they could be much much better, but I already overengineered this lol), improve the ui, make it more mobile responsive, and make it faster. I’ve also not submitted any pages to be indexed (technically any word/phrase in the English language could be a page.) Finally, there is marketing: I’ve done basically no marketing, and most of my visitors are through search (about 70 a day).

What should I focus on? My bandwidth is limited, so what should be my priority going forward? Learning as I go!

  • JouniFlemming@alien.topBEnglish
    1·
    1 year ago

    I think the main description line “Explore synonyms and antonyms of any word” doesn’t really describe this thing very well, considering it doesn’t mention anything about slang.

    Clicking the magnifying glass button or just hitting enter on the input field while it is empty results in 404 error message.

    But I think the main problem is that the quality of data doesn’t seem very high.

    For example, I typed in the word “man” and got a description of “someone who serves in the armed forces; a member of a military force” and some of the slang synonyms seem wrong, such as “sighs” and “MP”. And many other slang results are so odd, it’s difficult for me to know whether any of these are correct since some seem so obviously wrong. This creates mistrust and I think being able to trust the results of thesaurus, being that for slang or not, is a very high priority.

    I next tried your second most popular term, which is “home”. Its description seems okay, but some of the synonyms seem clearly wrong, such as “can” (which is repeated two times) and “.3”.

    I kind of like this idea, I just wish the data quality was a lot better. Right now, I couldn’t see myself using this for anything, as I would literally need to double check every data point from somewhere else.