If you use python it is very easy task to do (in python you can use Labelencoders from sklearn library and can apply CNN using pytorch or tensorflow anything) but as you are asking for implementing CNN in MATLAB

here are references:

https://www.mathworks.com/discovery/convolutional-neural-network-matlab.html

https://www.mathworks.com/discovery/text-mining.html

https://www.mathworks.com/help/textanalytics/ug/classify-text-data-using-deep-learning.html

The below Python framework is a repository for different deep learning based text classifications, and might be helpful

https://github.com/ddhruvkr/Deep-Learning-Text-Classification

Still I Have problem of how to eliminate missing values when dealing with large dataset ? How to managed data in prediction ( to predict values)