In this data-driven world, each and every byte of information generated holds enormous value to mankind. One of the common ways by which we derive value from data is to classify the data into different groups. For example, newspapers carry news items about sports, business, politics etc. in different sections which makes it easier for the readers. Similarly, in business context, oftentimes we need to classify production tickets for high, low or medium priority or we classify documents such as purchase orders, invoices etc. 

Often, this process is manual and time-consuming. What if there could be an AI system that does the job of classifying information neatly into various buckets? And, what if the AI system could classify any type of information without additional training? This is where Zero Shot Classification (ZSC) as a technique comes in. By harnessing the power of Natural Language Processing (NLP),  computers can classify information into categories without defining a set of crisp rules to follow in the process.

 

What is Zero-Shot classification (ZSC)? 

Zero-Shot classification is a problem in which the NLP model is provided with classes or categories that the model has never seen during training before. Even so, the NLP model will predict which class the given piece of text belongs to among the set of unseen classes. For example, if you give the text ‘I am happy about the weather today’ and ask the model to classify it as a positive or negative sentiment, the model would classify it as a positive statement. Similarly, for the text, ‘Please come home immediately’ and ask the model to classify based on urgency, it will correctly predict it as ‘urgent’. 

 Vuram’s ZSC Component:

Vuram has built and submitted a new zero-shot classification component in UI Path’s AI Fabric using Python and other open-source libraries. With the help of this, RPA developers can classify texts and other documents with a simple drag and drop!

Want to explore what Zero-Shot Classification can for for your enterprise? Write to us at ask@vuram.com 

 

 

COMMENTS

0 Comments

Submit a Comment

Your email address will not be published. Required fields are marked *

nine + nine =

Pin It on Pinterest

Share This