This is a series of Blogs that inspect the benefits Artificial Intelligence and Machine learning at the business level. The Technology landscape in this space is yet to consolidate, so the technology fitment is something that can be reviewed at PoC/Implementation stage.

In today’s blogs let’s take a deeper dive into the Print Media and Publishing Industry and how Artificial Intelligence and Machine learning can possibly solve some of the day to day challenges based on interactions with clients and prospects from this industry.

It’s a Goliath of documents/files out there – arranged in a precarious maze.  Not unusual to hear about billions and millions of documents, publications arranged systematically over network storage. Simple batch processes diligently performing their daily tasks over them. From what we usually see, it’s a mish-mash of IT processes that are untouched since they were first created with some gleamy hardware and lots of custom code.  No-one wants to rock the boat, probably for very good reasons.

Nothing could have pushed these printing factories into innovation, well almost – till cloud and mobility came. Very reluctantly the industry is now willing to polish their age old understanding of IT and come to realize that the world has indeed changed.

What we see are some ideal candidate use cases for artificial intelligence and machine learning.

  1. Classification

Billions and billions of units of data – documents, publications, files, contracts, articles – some structured, some semi-structured and some unstructured.  A Big Data Analysts’ mouth will water at the prospective challenge, but an almost certain win.  We have performed simple POCs to quickly attain a basic and functional classification engine that sorts the documents, learns their characteristics and improves its ability to classify.  All uncertain conditions are thrown up as exceptions, and each manual input further improves the efficiency of the system.

And talking about volumes, increased numbers simply increase the accuracy of the classification.

  1. Submission Management – Dealing with Multiple file formats

Publication houses receive submissions in ways all and sundry. There can be varying file formats – DOC, TXT, PDF, PNG, GIF. Another common disparity is the structures of the document – all authors write differently and some may well be incomplete writing.  So the content ingestion engine is a set of humans that open each and every submission, recognise the item, tag it appropriately and send for further approvals. Machine learning and AI driven system will eventually take over streamlining this process as well.  Machines can be trained to read submissions, understand what they are reading, tag them accurately and move them up the appropriate workflow paths. Not only is this process faster with the AI/ML tools, but also much more accurate. These systems can also be trained to read content from image files like GIF, JPEG etc.

  1. Sales Prediction Engines

The potential of prediction in publishing industry is tremendous.  There is plentiful data available on the internet and otherwise. Using an intelligent mix of social data, sales figures, author history and also select agencies a prediction engine can churn out the predicted sales figures of a title. This can help tremendously streamline the supply chain process.  The prediction can also be tuned to understand demand by geography and demography which means more optimized marketing budgets.  Having witnessed a prediction engine on the same lines, have seen these engines act as a decision support mechanismArtificial Intelligence Machine learning Publishing

  1. Reader Analytics

Reading habits steer you in the direction of the next bestsellers, and the benefit is in catching the trends ahead of times rather than being the last one at the party. This is an ideal scenario of prediction where excels will surely fail you.  Using the right data feeds and intelligent fine tuning of the business rules, you can understand what readers are reading more of. For example which genre is most discussed on social media in a certain geography, blend it with most commonly used tags on Instagram and intelligently understand the words that your readers understand and what they want to talk and hear about.

The possibilities are endless, feel like Alice in Wonderland.  The right way in AI/ML solution architecting is starting with Small POC steps and slowly building on accuracy.