Many publishers already use natural language processing to provide semantic enrichment and content recommendations for users. For example, using AI which can read and understand full text, publishers no longer need to rely on (limited and dated) metadata to identify their titles. At the conference Yewno, a vendor in this space, demonstrated their work with MIT Press using the Yewno tool to interrogate full text book content. The resulting analysis has thrown up significant new information about subjects and concepts specific to these titles which, in turn, informs editorial staff about new ways they can describe titles and relate titles to each other. The result – in this initial, limited test – is improved search results, relevancy and cross selling opportunities. Yewno and MIT plan to extend their collaboration into other segments of the editorial value chain including content acquisition and UI design. This work is particularly relevant to improving access to backlist titles, which often suffer from shallow descriptive and out of date metadata.
My other panelists at SSP (Storyfit, Elsevier, Unsilo, Molecular Connections) also described their activities using AI/ML to improve publishing workflows and discovery. Molecular Connections demonstrated how the American Institute of Physics (AIP) has used their technology to improve the experience for users seeking specific interrelated content. Unsilo is working with Taylor and Francis and the OECD to support better content discovery and collection development. Storyfit has a short video that shows how their technology works providing a good overview for potential customers.
Each of the AI/ML presenters at SSP demonstrated how artificial intelligence, sophisticated algorithms and deep analysis can be implemented to interrogate large datasets and corpus. This technology can also be highly leverageable to extend the capabilities of existing staff and to support, strengthen and expand the company’s product portfolio. The latter is especially true in the development of article and book collections.
Additionally, as an editorial tool these AI solutions can improve the accuracy of edits while also reducing revision cycles and cutting production costs. While we didn’t see that particular aspect of AI technology in these presentations, there is no doubt that the application of AI to the full editorial process will have a significant positive impact on workflow, staffing and cycle time. MIT's experience shows there is significant value to subjecting newly acquired titles to an AI filter to structure, add metadata and concepts and improve the submission before a human editor even looks at it. And it goes without saying that these activities will only take minutes to complete saving many hours of labor. Done correctly, the application of AI/ML tools can improve the overall productivity of existing staff.
The interest at SSP is just a hint of the general interest in AI and ML across all markets. A recent report by Pharus Advisors suggests that investment dollars are pouring into companies in the machine learning space. Here is their take:
As the Internet of Things expands and the amount of data being generated and collected continues to grow exponentially, Machine Learning is becoming a crucial part of managing and analyzing that data. Artificial intelligence can allow companies with high-volume data processes to vastly improve efficiency and productivity. However, growth in the AI / machine learning space has been slow until recently. McKinsey reported that in 2017 companies spent $39 billion on AI, three times more than in previous years. Furthermore, there has already been 70% growth in business value in AI during the first quarter of 2018, reaching $1.2 trillion (Forbes).
According to a recent survey, 61% of organizations most frequently picked machine learning / artificial intelligence as their company’s most significant data initiative for next year (Cloud-computing News). While market growth was slow for the past several years, 2018 is poised to be an explosive year for AI and machine learning investment.We are going to see and hear a lot more about AI/ML in publishing over the coming years, and many notable companies (Elsevier, AIP, Taylor & Francis) are investing aggressively to achieve cost and efficiency benefits, support improved customer experiences and develop new products. There are already enough examples and citations to make an informed decision about your strategy: If you are not on the AI ship soon, you will be left at the pier.