What is Named Entity Extraction?
In simpler terms Named Entity Extraction (NER) is a process of extracting entities from unstructured or raw text in Natural Language Processing.
The purpose is to classify or cluster documents based on the extracted entities. There can be multiple use cases of NER but at its core, it’s to classify & cluster or even identify the entities that it contains.
What are Entities?
Entities can also be referred to as topics. Entities are what Search Engines understand from the unstructured text content, the entities in that text content help them understand the content. Examples of Entities can be as follows
- Organization
- Person
- GPE (Geopolitical Entity)
Entity SEO focuses on the entity as the central element rather than making SEO all about keywords.
In a nutshell, it is important to signify entities to search engines, the entities that search engines are look for & finding it in top ranking documents.
Did the Google Algo Leak share any information about Entities with regards to SEO?
Wordlift did an amazing job at creating this Google Algo Leak Tool that gets you information on a topic & whether there was any information about that in the leak. Here I have extracted information about NER & we can clearly see that in the leak EntityType is all about that. Even in the example, they are talking about “person”, and “organization”.
Three Simple Ways to Extract Entities
1. Python Script through which you can extract Entities from shared text
Here is the Google Colab Python Notebook Link
Through this Python Colab Notebook, you will be able to visualize the extracted entities from the pasted text
P.S. This script can become far more powerful once we start leveraging Huggingface Open Source Models that are far more better at entity extraction
from the pasted text you can literally visualize the extracted entities. These are likely the entities that Google is being able to pick up as well.
Here are steps on how to use this script
Step 1 – Installations
# Install spaCy and download the English language model
!pip install -U spacy
!python -m spacy download en_core_web_sm
Step 2 – Import Necessary Libraries
# Import necessary libraries
import spacy
from spacy import displacy
from IPython.display import HTML, display
Step 3 – Load the English Model
# Load the English language model
nlp = spacy.load("en_core_web_sm")
Step 4 – Process the Text
# Process the text with spaCy NLP pipeline
doc = nlp("your text that you want to analyze & extract entities from.")
Step 5 – Visualize through HTML
# Collect HTML for visualization
html = displacy.render(doc, style="ent", page=True)
2. LLMs can also effortlessly extract the entities from the text
Using a prompt like this on Claude can help you not only extract entities but also visualize them within the text.
However, LLMs may have different classification categories. Categories that Search Engines may not support that is the reason I don’t place my trust entirely on LLM based entity extraction.
3. Google NLP Demo Tool
Here is Google’s own NLP tool in the demo of it you can paste the text & see the entities visualized.
How do NER/ Entities Tie into SEO?
Now coming to the part you were waiting for. How does Named Entity Recognition tie into SEO?
Here are a few ways
1. Is your content covering the entities the SERP Competitors cover?
This can be utilized in two ways, either you can paste the entire content for entity extraction comparing with competitor head to head to see the discrepancies OR you can just 1st introductory paragraph of you vs competitors to see the gap.
You will be shocked at times to see the correlation wherein the top 3 rankers have prominently covered a few entities & those entities could be altogether missing from your intro paragraph.
Semrush & other content optimization tools like SurferSEO, Frase covers this aspect in content optimization.
2. Entity Based Clustering or Tagging Content
Let’s suppose you have a giant publishing website that has a ton of content published but literally no contextual internal linking in place. It’s very random & manual. This is where you can cluster pages with their entities, tag the content with entities (i.e. creating tag pages) utilize tag pages into an internal linking logic & implement internal linking logic on the basis of it.
Kunjal Chawhan founder of Decode Digital Market, a Digital Marketer by profession, and a Digital Marketing Niche Blogger by passion, here to share my knowledge