Home > Natural Language Processing
Search engines are becoming more capable of understanding searcher’s intent as a result of development in the field of artificial intelligence, or more specifically natural language processing (NLP).
The early days of SEO were very different, Google’s ranking algorithm was far less advanced. And as a result crawlers simply looked at specific keywords to infer what was the meaning and relevance of the page.
Due to the advancement of AI technology and Google’s ranking algorithm this is no longer the case. For example, the development of Google’s Knowledge Graph and the Hummingbird core algorithm update. The knowledge graph allows Google’s algorithm to better understand the relationship between particular entities and concepts, and the Hummingbird update helps Google better understand the meaning and context behind queries.
Major impact from the Google BERT and MUM Update
The Google BERT and MUM update as well as other NLP technologies have had a significant impact on the SEO industry.
Google’s Bidirectional Encoder Representations from Transformers (BERT) algorithm update in 2019 transformed both NLP technology and the SEO industry. The BERT algorithm understands context and nuances of words in search strings and matches those searches with results closer to the user’s intent.
Google’s Multitask Unified Model (MUM) algorithm update further expands upon the BERT model and when it is implemented in late 2022 it is expected to revolutionise SEO and how users use the Google search engine.
The effect Google’s MUM’s algorithm update will have on SEO is further expanded upon here.
What is Natural Language Processing?
Natural language processing (NLP) is a branch of computer science that studies how to give computer systems a better understanding of text and voice data to a similar extent as humans. This is achieved by combining computational linguistics with statistical, machine learning and deep learning models to allow computers the ability to understand the intent and sentiment expressed by a writer or speaker.
Advancement in artificial intelligence has allowed NLP technology to become much more commonplace. It is highly likely that you have interacted with such technology today; whether that be a voice operated GPS system, digital assistants, customer service chatbots and/or the case discussed within this post search engines.
What is Semantic SEO?
As search engines have become more capable of understanding the world the same results can now be presented to users across different search queries if the same user intent is perceived by the search engine. With linguistic AI and word-vectors, we can start exploring a concept to see how it is semantically related to other concepts. This has moved some of the attention away from targeting specific keywords towards becoming relevant for a specific topic cluster.
Semantic SEO is the practice of creating search engine optimised content around topics, and not individual keywords. There are a number of different effective methods towards building topical relevance; one such method is to expand the topic in all the directions that may be of interest to the user/your target audience.
This allows Google crawlers to better understand your content and also help them view your content as high quality and deserves to be promoted more often on the SERP.
How NLP can help improve SEO
Ensuring that your site accounts for Google’s intentions of giving NLP a larger role in determining search rankings in the future with the Google MUM update will be essential moving forward as a means of conducting effective SEO.
There are a number of ways NLP technology is already used to improve SEO and user engagement that may be expanded upon in the future as technology advances; some of these include:
Structured Data Markup
Structured data markup is very important in increasing organic search traffic and overall SERP ranking. The markup that is added to your HTML helps search engines to better understand and process your content and makes your site eligible for rich results, like image and video carousel and knowledge panels.
The structured data markup describes your content using entities; a word or phrase that represents an object that can be recognised and categorised. These objects include: people, events, consumer goods and organisations. Google is able to distinguish between these entities and use the information to better search results to satisfy user intent.
Internal Link Building and Content Discovery
Natural language processing algorithms are often trained on large semantical databases like Wikipedia using machine learning techniques. They rely on internal links to provide additional information on what the content is about and how to help users discover relevant content from your website.
The benefits of effectively implementing an internal linking structure on your site are numerous; for example, it can significantly improve user experience by ensuring that your site is much more navigable for new users and web crawlers and in turn improve the SERP ranking and overall SEO health of your site.
There are many machine learning software tools available on the market today that can crawl all the pages on a website and discover opportunities to include an internal link or restructure/change anchor text to include relevant keywords that may increase click through rate or find broken links. They achieve this by analysing and identifying trends within the choice of keywords and links already included on the page and site as a whole.
For a further explanation on the importance of an effective internal linking structure click here.
Content Recommendation
A content recommendation system is a type of machine learning algorithm that provides users with relevant recommendations. This algorithm is usually used as a means of improving user experience and dwell time, the time that users spend on the website between the initial click and the return to the result page.
The better the recommendations the more engaged readers are with the content and the more likely they are to convert, share posts on social media or link to the content on their own sites.
NLP technology is able to achieve this by using machine learning models to classify information and make predictions about what the user will like to read next. Metadata is stored in machine readable formats like JSON-LD or a Resource Description Framework (RDF); adding a semantic layer to the content’s metadata can greatly improve the machine learning models capabilities as seen in some of the more advanced content recommendations systems on the market.
Spotify perhaps has one of the most advanced content recommendations systems currently on the music streaming market. If a user is a fan of rock music and listens to rock music every day, Spotify will begin to recommend more rock music and music of related genres. They curate these songs and create custom playlists for users that increase their switching costs of moving to another music streaming platform as user data is only held on that platform and will rarely transfer to accounts on competitors platforms.
Smart Redirections and 404s handling
Smart redirections are a mechanism that allows a website like in this case Quora to route users to the right topic by expanding the synonyms of a given topic. This allows users to find web pages much easier
For example, below you can see the browser automatically redirects the request to a topic page for ‘Search Engine Optimization’ on Quora:
The Quora web server has been built to understand that the term ‘SEO’ is synonymous with ‘Search Engine Optimization’ by using knowledge graphs like DBpedia where all of the synonyms for a given concept are described.
This allows Quora to be able to configure multiple 301 redirects to intercept requests without having to worry about how each user is calling a specific concept.
Although the inclusion of NLP technology within Google’s core algorithm updates may necessitate marketers to make some changes; it is unlikely to have any significant impact on already well established SEO principles such as the importance of page speed, link building and producing high quality content.
With the exception of some algorithmic changes you will likely only need to start including more long tail keywords in your content to provide more contextual information to the Google crawlers to accommodate for the inclusion of NLP technology within the SEO industry.