Hi, my name is Fogang Fokoa, PhD student in machine learning. I have been a data scientist for more than 2 years.

This service aims to solve your NLP problems using neural networks and python. Mainly, this service will be able to solve 4 types of NLP problem: text classification, sentiment analysis, topic extraction and question answering.

The solving methodology will be based on the CRoss Industry Standard Process for Data Mining (CRISP-DM) which has 6 steps:

(STEP1) Understand the problem. Understand the objectives and requirements of the project, select the technologies used; examine data, its properties such as data format and number of records or field identities.

(STEP2) Understanding the data. Exploratory Data Analysis (EDA): draw graphs and analyze them;

(STEP3) Prepare the data. Clean, select, format, engineer and transform features;

(STEP4) Modeling. Generate a test design, perform a spot-check then select one model to solve an NLP problem (text classification, sentiment analysis, topic extraction, or question answering). The models (just one model will be chosen) that could be used are:
--- CNN;
--- LSTM;
--- seq2seq;
--- Word Embedding;
--- PCA, SVD, Latent Dirichlet allocation

(STEP5) Evaluation. Scoring model and review (1)-(4);

(STEP6) Deployment. Produce a final report, plan monitoring and maintenance, make the models operational;

The solution will be delivered as a jupyter or .py notebook depending on your needs.

**************************** Basic Formula *****************************
--- Debugging help on a Python Deep Learning script
(STEP1) Understand the problem
(STEP2) Understanding the data. Data visualization


******************************* Option 1 ************ ********************
(STEP1) Understand the problem
(STEP2) Understanding the data
(STEP3) Prepare data


**************** Option 2: TEXT CLASSIFICATION ****************
Add to Option (1) the (STEP4) and the (STEP5)
(STEP4) Modeling. Dimension reduction could be used in any of the model types depending on the problem encountered
(STEP5) Evaluation


**************** Option 3: SENTIMENT ANALYSIS ****************
Add to Option (1) the (STEP4) and the (STEP5)
(STEP4) Modeling. Dimension reduction could be used in any of the model types depending on the problem encountered
(STEP5) Evaluation


**************** Option 4: TOPIC EXTRACTION ****************
Add to Option (1) the (STEP4) and the (STEP5)
(STEP4) Modeling. Dimension reduction could be used in any of the model types depending on the problem encountered
(STEP5) Evaluation


**************** Option 5: QUESTION ANSWERING ****************
Add to Option (1) the (STEP4) and the (STEP5)
(STEP4) Modeling. Dimension reduction could be used in any of the model types depending on the problem encountered
(STEP5) Evaluation


**************** Option 6: WRITE A REPORT ****************
Add to Option (2), Option (3), Option (4) or Option (5) the (STEP6)
(STEP6) Deployment. Produce a final report.


Thank you.