Smart Content Authoring

Rishiraj Saha Roy

Harsh Jhamtani

Adobe Research

Natwar Modani

Adobe Research

Niyati Chhaya

Adobe Research


Enterprise content writers often need to repurpose existing content for a different need. This calls for an algorithm to automatically understand the content writers’ context and use this to identify previously written content from the same enterprise to be repurposed. Our initial set of algorithms in this work focuses on automatically constructing a query from an author’s content to be used to retrieve relevance content for repurposing. Once the content is created, it is also important to modify the content to suit the needs of the channels that it is delivered to and content writers are also expected to deliver such channel-modified versions at a rapid pace along with numerous other stylistic personalizations of the same piece of content. Examples of such personalizations include short textual representations for social media and verbose representations for emails, newsletters or websites. To maintain high production rate, automated content modification techniques are desired. In this work, we also explore algorithms that take a piece of textual content as input and summarize/expand it to a desired size. The input content is a short snippet composed by the author along with a desired size of the content as required by the channel. Once the variations are created, we also have technologies that can efficiently manage these variations. We also propose algorithms to automatically seamlessly propagate the changes in a master content to its several variants.

Project Publications

A Supervised approach for Text Illustration

Jhamtani, H. (Oct. 15, 2016)
MM 2016 - ACM Multimedia Conference

Generating Multiple Diverse Summaries

Modani, N., Srinivasan, B., Jhamtani, H. (Nov. 9, 2016)
17th International Conference on Web Information Systems Engineering