/ Information Consumption / Education /


Create a way to reduce a book’s word count to 1.5% whilst improving comprehension

contact lens AR overlay


1. Wear contact lens with internet connection and AR overlay.
2. As you look at book, AI would scan the content from Google Books or Kindle and extract most highlighted content on Kindle.
3. Search web for book title and phrases from most highlighted content.
4. Determine from search results what action people take after reading content.
5. Present results as an overlay with 2 most impactful actions beside it.
6. Everything processed in 1sec or less.

Why not reduce it by more than 50 % through summarisation technique


Update: I miss-read the topic. Please ignore this. Thanks. 🙂

Reducing only 1.5% of the total word count, will only reduce the page count of a book by 10 to 15 pages of a book.
For a novel the number of page count is 60000 – 80000,and no more than 100,000. Let’s say the average word per page is 300 words (which normally is 250-300 & depends upon font size and arrangement of words).
Now, the number of page becomes = 90,000/300 pages = 300 pages.

Let’s say we reduce the word count by 1.5%.
So now the no of words becomes = 90,000×98.5/100 = 88650 words.
So now the book will have a minimum = 88650/300 = 295.5 pages

Therefore. we still need to read 295 pages of the book to get the idea.
So, why not summarise the book.
Here, we are strictly talking about non fiction books. Every non fiction book is filled up with ideas or knowledge or theories of the author. The author then gives his or her ideas a shape by providing real facts , general truths , reasonings and logical conclusions. Finally he or she adds her opinion to it.
To me every chapter has a topic, a background , a central point , a fixed scope of the topic and finally an opinion. Therefore, if you read, every chapter inevitably falls into a background- central point – evidence model.
If we can figure out this model, we can summarise it by taking a small portion of the background and evidence and putting it with the central idea of the chapter.
Backgrounds are always a past fact that has taken place and sometimes a general truth. Similarly evidences that the author present will be examples, which can be further classified to facts. Both evidences and background can therefore be separated from normal judgement/ opinion/ conclusion.
This kind of summarisation technique has been already done. But doing it with a software will require making AI models as described above.
AI models cab then be trained by ranking the summaries.

Mimic human


The crawler over each text in the book. The crawled text are used by the AI-bot-1 to make a diagram/model (preferably a data structure (In simple, like a directed graph) ), make all the nouns as objects and the verbs as interface between objects. Seamlessly crawl the online for reviews, short notes with the book name. Add each with a start and end point in the diagram; this reduces a long text with crisp words.

Other, AI-bot-2 to find the pattern in the diagram with the knowledge of the previous reads, that helps to give references to other things; this reduces the need of much text.

While crawling over texts, find for twists (an emotional change, a dramatic twist – by looking for words suddenly… ) and give a priority stamp to those blocks in the model. Look for time references and give a time-stamp with relevant to other events in the model, and sort the model with respect to the time. considering initial time for all events as zero.

Apply algorithm, to remove dead ends (that maybe a filler, or a short story within it). Dead ends as the objects will have no further connections (mostly, events make a connection) to any other objects. Also loops, as maybe a daily routine. Importantly, must consider the priority of these too and take a decision.

In the diagram/model find the starting point(may not be the starting of the book) and ending point (climax); then find the path with most priority. Before it, sort the model with time.

Convert the model into text consisting the main path and other high priority ones. Use the reviews and comments to replace the start and end point and use the references to other books/texts; this will make more humanly.

For conversion, the AI must be trained for. And also uses its own previous knowledge.

1) Create a model/datastructure to store the flow of the story.
2) AI to build the model
3) Find time of events and the priority
4) Find the start and the climate
5) algorithm to find the more priority path from start to end
6) crawl internet for references, reviews and comments
7) AI to convert the model to text with references, reviews, comments and its previous experience.

*In general I mentioned it as AI, it can be machine learning, deep learning or nlp. The most suitable one for the task.

Reduce Word count and increase comprehension using Natural Language Processing Techniques


1.) Define category of book’s content – Fiction, non fiction, politics etc
2.) Compare the content with a corpus of other books in the same category
3.) Every book has characters, organizations, buildings, etc, a canvas. Understand the named entities in the book
4.) Using natural language processing list out events surrounding the name entities
5.) Build relations between the events and relationships between the named entities
6.) Summary the named entities, relatationships between named entities, events that occur
7.) Using the corpus one can also eliminate text that can be ignored as it is only explanatory of surroundings etc
8.) Tweak it multiple times to get to 1.5% word count

Zero Font


I am sure working on a new font will solve this issue. A font that is designed and reengineered to use custom space than required.

Submit your own idea!

Submit your idea to earn a future revenue share!

+ Submit Idea