kfarias / complete-me Goto Github PK

Application is broken into components which are well tested in both isolation and integration using appropriate data

The common gripe about autocomplete systems is that they give us suggestions that are technically valid but not at all what we wanted.

A solution to this problem is to “train” the completion dictionary over time based on the user’s actual selections. So, if a user consistently selects “pizza” in response to completions for “pizz”, it probably makes sense to recommend that as their first suggestion.

Your library should support a select method which takes a substring and the selected suggestion. You will need to record this selection in your trie and use it to influence future selections to make.

Here’s what that interaction model should look like:

Once the words are placed into the trie it should also offer some suggestions.

Application meets all requirements, and implements one extension properly.

Our Trie won’t be very useful without a good dataset to populate it. Our computers ship with a special file containing a list of standard dictionary words. It lives at /usr/share/dict/words

Using the unix utility wc (word count), we can see that the file contains 235886 words:

See if you can implement a front facing application for your trie. The user should be able to submit a word and then receive the suggestions on the dom.

Application is expertly divided into logical components each with a clear, single responsibility

Application demonstrates excellent knowledge of JavaScript syntax, style, and refactoring

The first thing your trie should be able to do is take in a word.

Application consistently makes use of the best-choice Enumerable methods