bfbachmann / silver-screen Goto Github PK

We should either set up default params on these models or handle this exception on model.save()

I'm working on a fix now.

I occasionally find that I encounter "Twitter API is down" errors when I'm developing. I believe these to be caused by errors during the processing stage (which are resolved through development), but they aren't related to the Twitter API. We should make sure that the "Twitter API is down" message is only shown for API errors, not otherwise.

We need to implement a proper overall score generator once we're confident in the sentiment score generation.

If we have a tweet in our database, we are currently tying it to a single movie. In cases where two movies use the same tweet (i.e. a franchise with simplified search terms) this causes no tweets to display alongside the other movies.

We have received a lot of feedback asking why we aren't able to search for Tv shows, so adding the word "movie" to the subheading might make it a little clearer.

clearly define that you can only search movies :(

Some elements in the navbar have margins that are too large and cause it to expand.

If a tweet has been re-tweeted, this should impact the sentiment score somehow

If a twitter user rates a movie in their tweet (AKA "Finding Nemo was pretty good, 4/5"), our algorithm should heavily weight the score they provided ("4/5") when determining the sentiment score.

Implement an auto correction during the text parsing phase of the sentiment analysis, so we can get scores from misspelt words.

Since we added more search term restrictions (i.e. adding 'movie' to the query if it's a common phrase) we're more likely to find nothing on a random search. If we do find nothing, we should be searching again until we find a movie that does have tweets associated with it.

Could throw a message that the search term was not matched exactly and the most popular related movie has been returned or put in an about page that explains the returned movie

We should probably change this to a redirect to /index

There is currently no error checking for when we pull OMDB data.

I'm in the process of writing an automatic job to purge trending items daily, but haven't made much progress. It should be done by beta release 2.0.

Using additional movie related hashtags to search for movies. Could help to resolve issues where if people are tagging their movies with slightly different tags (such as hp for harry potter) we can find these tweets as well.

If the user dismisses a trending notification (by clicking the 'x' button), it should not return until trending status is updated the next day.

If a movie contains only words that are extremely common, then add "movie" to the end to clarify the search parameter

We should instantiate it only once for performance reasons.

Valence, etc

Change way lexicon class is handled so we create an object instead of using only a text file

We need to attach a postgresql database.

I think the website might look a little better if we centred "Twitter Sentiment Analysis Tool" with respect to the title. This obviously isn't an essential fix, but I think it would definitely make it look more professional.

When making a GET request to Twitter we never make sure the the response is actually what we expected. We should be checking for 500 or rate-limit exceeded (88) status codes, and probably a few more. These errors also need to be responded to on the front end. We cannot leave the user hanging.

Cannot find anything on twitter when movies have overly complicated titles. Presumably because people do not tag their tweets with extremely long titles ex "Lord of the Rings: The Fellowship of the Ring".

Make it easier to browse through the code and find classes quickly

I encounter this error each time a tweet is put into the database, flooding my console. It doesn't ultimately seem to impact the storing of tweets, however.

The chart looks like this on Safari if one searches for a movie:

Currently our sentiment analysis algorithm takes the name of the movie into account when analyzing the sentiment of each tweet. It should filter out the title because of the effect the title could have on the sentiment score for that tweet. Here is an example of where this ruins our data: