Claiming I will create the best page grader and actually defining it are two very different things. Luckily, there are plenty of natural language processing tasks that can empower an SEO or content copywriter. Here is an overview of features we may look to implement.
Text Statistics
Text statistics is mildly useful, but can be helpful to benchmark how well a topic was covered. What's great about it is how well you can see content growth over time.
Keyword Extraction
Keyword extraction uses the idea of centrality to find the most central phrases in your content. Compare your central keywords to your target keywords to understand how well you covered a topic.
Entity Extraction
Google uses entities to power it's knowledge graph. If a search term triggers a knowledge graph panel in Google, then that means the ranking content is using an entity as a keyword.
Text Summarization
Google uses text summarization to power their meta description replacement. If you happen to find that your meta descriptions are being replaced, it is likely because your meta description doesn't properly describe the searchers intent or the content on your page. This will allow you to compare similarity between the meta description and an automatically generated one.
Internal Link Analytics
Internal links are a powerful lever. It's like being able to point at a piece of content and tell search engines it's important and related. Calculate Internal PageRank to see which pages you think are important.
Question & Answer
NLP isn't great at finding questions, but it's very good at finding answers. With some pre-seeded questions, we can see how well your content answers the queries of site visitors.
Sentiment Analysis
Not so useful. It's either positive, negative, or neutral. Sometimes a mixture. It has it's place in NLP, but probably not in Lemmatic.
Semantic Search
Given a keyword, we can see how well all content scores up to that keyword on a scale of 0-100%.
Keyword Position
Many old-school ranking systems rely on keyword placement in the title, meta-description, and first 100. It's not so relevant today, but I believe it's still a best practice.
Text Classification
Text classification is handy for picking a category automatically. It may have a place for automatically categorizing content, but not for this tool.
Topic Clustering
Topic clustering is similar to text classification. It could suggest new categories, but it can also often be wrong.
Content Similarity
Content similarity is perhaps one of the most powerful tools. Lemmatic could suggest either internal links, or if the content similarity is extremely high, maybe you want to rewrite, update, and merge the two signals from the separate posts.
Natural Language Generation
GPT products from openAI have taken the world by storm. It would be powerful to connect natural language generation directly to the CMS to help content editorial move and iterate faster.
Overall, there's huge amounts of room for tooling improvement that I can bring directly to a CMS market.