Writer's Content Complexity Measure Just Got A Whole Lot Smarter

Posted on July 06, 2016
By Suhash

We recently pushed out a new update to our platform. You probably didn’t even notice – but look closer and you’ll see it immediately. 

The update addresses various issues but mostly focuses on a massive overhaul of our Content Complexity measure.


 

Content complexity provides you with synonyms based on what audience knowledge level you wish to write for.

For example: if you primarily write articles for industry professionals in a very technical field it is understandable that your piece contains jargon and industry-specific terms.

In contrast an article written for a more general audience will be written for industry newbies or those looking to learn more. Here, you would not want to use as much jargon or technical terms because it may go over the reader’s head.

Furthermore, writing at a very basic level for industry professionals may signal a lack of credibility to the reader. They may disengage with your post and look for information elsewhere that targets their level of knowledge.

That’s where our content complexity measure comes in. It pinpoints a word that does not fit with your desired audience level and then provides you with more suitable synonyms. It’s like an intelligent thesaurus and has the ability to give you simpler or more complex synonyms.

Several clients had mentioned on multiple occasions that our content complexity measure was a bit off. Word replacement suggestions were extremely limited – and those that were suggested tended to be out of context and not applicable to the context of the phrase. Suggested replacement words did not match the desired readability level – missing the mark when you tried to improve the article’s Atomic Score.

This update aims to address these issues – it involves tweaking current mechanisms and adding new ones to deliver a better overall user experience.

Let’s break down the Content Complexity measure further.

How we use readability tests to offer accurate word recommendations

Our platform uses a highly accurate readability test formula as the base for all analyses. The test determines the complexity of any given amount of text. Using the test formula as our baseline measure we created our 5 readability levels; General, Knowledgeable, Specialist, Academic and Genius.

By incorporating this formula into our measures, we are able to offer more accurate word recommendations that are more in-line with the desired readability level that you want to achieve.

Upgrading our parser for better part-of-speech detection

Parsing is an integral part of the platform. In a nutshell, a parser, in this context, takes a sentence and tells the computer what part-of-speech each word is. The old parser we used was not doing a great job – it would mislabel the part of speech of a particular word and therefore provide inaccurate recommendations for replacement.

Our new parser is far more advanced than our previous one and delivers more accurate results.  We are now able to accurately determine which part of speech a particular word is and how it fits within the context of a sentence.

Implementing a new semantic network to deliver accurate synonyms

Our final upgrade is out of a need for better synonym recommendations – specifically synonyms of particularly ambiguous words, or words that have several meanings.

Here’s a simple example: The acting Director of Finance enjoyed visiting his cottage by the lake.

The word we want to pay attention to is “acting”. Our previous database would recommend words related to acting, in the theater/movie context. Now, it is intelligent enough to figure out that acting here refers to words such as temporary, deputy or interim.

The semantic network we integrated into the platform is a database that consists of pairs of words, and the relationships between them. 

Let’s look at the relationship between a guitar and rock music. Put it in the context of “Used For” -where a guitar is used for rock music. In a nutshell, it gives rules to the engine on how to interpret common-sense relationships between objects, providing more accurate synonym recommendations.

Improved for a faster and more accurate experience

The new content complexity feature is a major step up from its previous iteration. By incorporating excellent open-source resources into our platform we ensured that synonym recommendations are as accurate as possible.

Give our improved content complexity measure a shot to see what it delivers, and share any feedback that you may have.

Happy Writing!

About the Author:

Suhash is part of the BizDev team @ AR. When he isn’t at work, Suhash can usually be found parkouring across Toronto with a crossbow. That, or reading a book.

AR Blog CTA-final-01

Tags: software, wordpress plugin, writing, writing tech, App, app update, Atomic Reach News, AtomicWriter, blog, blog writing, blog writing tips, blogging, chrome extension, content, content creation, content marketing, Content Marketing, Machine Learning, marketing technology, martech, saas, Technology, web app, writer, writing technology, writing tips