alahnala / iMessageClassifier

We makin' some kewl stuff.

Mac application

Requirements

• macOS 10.11 or later
• Python 2 and nltk package
• (Optionally) Xcode 8.1 to compile in Xcode

Files

Source used for the Mac application are in the app directory. The compiled application is in the root directory for the project, called MessagesSentiments.app.

Running Mac application:

• Option 1: launch compiled app:

  • Double-click on MessagesSentiments.app

• Option 2: lauch from Xcode 8.1:

  • Open app/MessagesTestMac/MessagesTestMac.xcodeproj in Xcode
  • Set scheme to MessagesTestMac > My Mac (top left corner near Run and Stop buttons)
  • Run

• Set Python Path to the path to Python 2 on the computer.

  • The path can be found by executing which python command in Terminal.
  • You also must have nltk package installed, as well as the corpus data it uses.

• Open chat database.

  • The sample database is provided in datasets/chat.db.
  • Actual iMessage database for the user is located at ~/Libarary/Messages/chat.db.

Python classifier

Requirements

• Python 2 and nltk package

Files

Source files used for the the evaluation are in the root directory for the project, specifically feature_vector.py, process_training.py and parse_train.py. In addition, the classifier requires a JSON file with messagese, an example of which is found in datasetes/messages.json and a labels file, as well as Sentiment Analysis Dataset.csv and unigrams.txt, both of which are in the root directory for the project.

Running feature_vector.py

• Execute the following command in Terminal:

$ python feature_vector.py messages.json "Sentiment Analysis Dataset.csv" unigrams.txt

Evaluation

Requirements

• Python 2 and nltk package

Files

Source file used for the the evaluation is in the root directory for the project, called get_accuracy.py. In addition, evaluation requires a predictions file, an example of which is found in datasetes/labeled_messages.json and a labels file, an example of which is found in datasets/xinrui_labels.txt.

Running get_accuracy.py

• Execute the following command in Terminal:

$ python get_accuracy.py <predictions file> <labels file> 

Datasets

Included in the datasets directory are

• chat.db, a sample iMessages database that can be used in the Mac application.
• messages.json, a sample JSON file of messages created by the Mac application and used by the Python classifier.
• labeled_messages.json, a sample JSON file of sentiment labels created by the Python classifier and used by the Mac application.
• xinrui_labels.txt, a text file with sentiment labels created by humans, used for the evaluation.