You have some text and you need to classify it. Your existing team does not have the natural language processing expertise to build a good model, or you need to exceed that accuracy that your team can get with off-the-shelf tools.
At Electric Brain, we can build text classifiers using a variety of methods, both with and without deep-learning. This is a simple and straightforward use-case of machine learning technology, and our expertise with natural language processing makes it quick and simple.
We assume in this project that we can get access to the data without needing to build a lot of hand-baked transformation scripts.
Time & Cost
You provide us a spreadsheet or CSV file containing all of the text, along with the labels they need to be classified as. No data wrangling necessary, we’re just doing the core modeling.
Time: 2-3 weeks
We have to directly access the database to get the data. It’s not in a convenient form, so we have to write our own data-wrangling scripts. But the size is manageable.
Time: 3-4 weeks
You have too much data than can be stored on a single computer - the data is spread out over a cluster of database servers. Special care must be taken in the handling and processing of this data, because of the vast quantity available.
Time: 5-6 weeks