Classification of e-mail messages in Theme folders
$30-250 USD
Paid on delivery
1 Prepare a program that will classify e-mail documents in thematic folders.
2 The program should operate on e-mail files. You can save them on
drive and operate as text files. You can also write a program in
a plug-in to your email client (eg. Thunderbird)
3 Prepare a training set size of at least 100 documents and a set of
Test the size of a minimum of 50 documents. You can use the message
own mailbox or generate them. In both collections should be found
documents from all selected topics.
4 Select any three topics that you want to classify messages.
Manually mark (in both sets) documents that belong to each class.
There also may remain untagged documents.
5 Choose 2 any method of classification(Multinomial Naive Bayes, Bernoulli Naive Bayes), the program should allow you to select
classification method for the assignment of documents to folders topics. On
a set of training program should build two classifiers.
6 classifiers should use information about:
- The message
- The subject line
- Sender of the message
- The date of dispatch
7 Check the percentage of correctness of the classifiers on the test set.
8 The program should allow an arbitrary check new posts
e-mail.
Project ID: #6288176
About the project
5 freelancers are bidding on average $303 for this job
I am willing to discuss further about the project specifications and deliver the same to your needs.
hello, i am red hat certified engineer and i easily can do this task done i am more then 4 year experience in this field, i need a chance to prove my skills to you i am ready to start now thanks