Text classification is the smart categorization of any text corpus into various predefined categories and is one of the most common ways to structure unstructured data to analyze it. In one of our previous blog, we discussed different machine learning techniques to classify text including our latest offering – Custom Classifier. In another article, we discussed how you can use Custom Classifier to build your own text classification model on your custom defined categories without any training data. In this article, we will see how custom classifier can be used for explicit and implicit text classification.
Explicit Text Classification
In order to better understand what we mean when we say Explicit Text Classification, consider a text input to assign an emotion to it, “The coach was disgusted with the team’s performance” with the following categories to classify the input text into: Disgust, Happy, Scared, Sad. What would be your wild guess? Yes, It is disgust.
This is an exact example of an explicit text classification, where the input text either carries the categorical classification or in itself is directly pointing towards the classification that you have to assign to it. Let us see how our custom classifier performs in such a scenario through the following example.
Implicit Text Classification
Implicit text classification can be thought of as classifying text into categories without the mention of any direct relationship between the text and categories defined. For example, if you think of Lionel Messi what instantly comes to our mind? Soccer ?. But when we talk of an automated text classification model that has no context of Messi’s background, would that be able to identify the same example in a similar manner? If you are talking about our custom classifier then the answer is a whopping Yes!
Don’t believe us? Try our demo!!
Custom classifier was able to classify that Lionel Messi is not just another name but also related to Soccer. This was an example of what is an implicit text classification wherein the input text there is no direct indication of or to the categories that it needs to be assigned to.
Not convinced yet? Let us check out another cool example from the recent hype in the news regarding the transfer of Alexis Sanchez to Manchester United.
Isn’t that great? Our classifier picked up the league without even any mention of the sport !!
As we can see, this is among the first budding steps towards general AI. We did not mention anything related to Premier League or soccer but our Custom classifier could identify the relation. The very ability to do an implicit text classification over explicit text classification is what makes the Custom Classifier an amazing and useful tool.
This was a far-fetched concept until recently. As we know, the biggest obstacle that the usage of Machine Learning and Text Classification projects has faced so far is the ready availability of an annotated dataset to train the algorithms. But with the ParallelDots’ in-house research team’s recent zero-shot learning inspired algorithm behind Custom Classifier, we are truly living upto our vision of providing AI at your fingertips.
What does this mean?
This clearly states that in action our Custom Classifier stands out of the league by miles! Imagine, now you don’t need a data scientist or an ML expert to make a text classifier for you. Also, as you can see that we give categories which are defined by us right at the point when we are carrying out the analysis, hence, there is no pre-training required to carry out text classification so you don’t need a training dataset customized for your classification.
This empowers you to carry out text classification without having to worry about the kind or domain of the input. So you can actually carry out say both classification of legal news and sports news using the same tool.
Why just that, this means that irrespective of any kind of data, be it tweets, facebook comments, news or reviews, you will be able to carry out your text classification task without ever having to worry about anything else. Also, being able to carry out both explicit and implicit classifications you can build a smarter search engine for your content or make your conversation bot more intelligent.
Want to try it now? Signup for a free ParallelDots AI APIs account to get started.
Custom Classifier is a leading example of how AI is going to shape our future Natural Language Processing needs. It not only gives you an edge over any other text classification tool by letting you define your own categories on your own input data but also makes your text classification take into account any implicit or indirect relation with the categories that you define.
The very ability to carry out both explicit and implicit text classification simultaneously gives you an incomparable edge in carrying out any text classification. Custom classifier thus becomes the pioneer in helping you break away from any shackles that had kept you from using the real power of text classification.