Terminal Search

Terminal Search is a project where it started with crawling the web to find the news headlines and clean the data to learn sentiment analysis.

This project consist of three seprate functions that operate sepratly form one another, which I will explain further later.

JavaScript/React/MUI and Python are used for the front-end and the back-end code development. For handling communication between front-end and back-end I used Flask and Axios. You can view the source codes in repository here!

The header of the page consist of three diferent functionalities within the app.

Initial page after running the program and having different option on the header.

The first tab consist of News Sentiment Processor. In this section of the page, user can search for any prompt they wish to and process the sentiment of the searched promt in the headlines of news. Users can see if the sentiment is positive or negtaive based on 3 different trained models: NLTK Sentiment, Text Blob Sentiment and Vader Sentiment.

Users can choose to get ther result of search propmts sentiment in all titles as a whole, or get to see the result of sentiment for each title alone and separatly. 

After searching the prompt user can see result of all titles sentiment.

Prompt sentiment analysis of titles sepratly and giving it a score at the end.

The other tab allows user to check for a text to see if it is a food recipe or not.

Users can type, drag and drop, or upload a file of text to check for the information being a recipe or not.

The 3 models that I have used for checking the data are Adaboost, SVM and Random Forest.

Text process result to check if the input was data of food recipe or not.

The final tab consist of word prediction, in which it is supposed to prompt and replicate a search engine such as google, which suggest words while users type in the search propmpt. 

The Model is trained on texts within a book, and as user types in the text, the input will be sent to the backend and the last 4 index words of the sentence will be sent for process and prediction. 

The prediction is indicated under the search bar.

Since it is trained on a text book, it doesn't predict and produce the best predictions but does it job!