How the Semantic Web and Machine Learning Are Changing SEO
The world of search and search engines has been changing and changing fast. We are moving from typing keywords on search engines like Google, Bing, and Yahoo to voice search, which conversational search. In fact, to get information these days, you do not even need to use the traditional search engines as we know them, but rather what is known as an "interactive agent". (Don’t worry, we’ll explain that term in a minute.)
Before I lose you with all the terms, it might help to explain that all these concepts are geared towards one goal: serving better, relevant results, faster. It might help further to understand that this goal originates from a concept known as the semantic web.
The semantic web, as you could guess from its meaning, is geared towards meaning. It aims at getting to the user’s search intent so that responses served are the correct ones.
In this article, we will focus on an aspect of the semantic web that allows you to "talk back" to search engines, and how websites can contribute to serving relevant results.
Let’s dive in.
Background of Semantic Web
We will start with a little background. The possibility of "talking back" to search engines was first heard in 2013 during a Google I/O developer conference. This concept involved more than voice search. The computer was able to speak back to you when giving answers to your questions. Better still, it was able to answer follow up questions based on the same subject. This was possible if you were using Google Chrome.
Let’s look at some examples: If you asked, "How old is Taylor Swift?" Google was able to give you an answer. If you then asked, “does she have siblings?” Google was able to give the answer to that question, too. This was, of course, an exciting technology. This concept of “conversational search” broke new ground for more advanced forms of semantic search that we will look at next.
Advanced Forms of Semantic Search
1. Google Now
If you have heard or used Amazon’s Alexa or Google Home, it will be easier for you to understand Google Now. If you haven’t, not a problem, as we will delve into it now. Google Now is what we can refer to as an interactive agent, intelligent agent or a context engine. Google Now is able to understand user intent — what you are really searching for — and give you relevant results.
How does it do this? It is able to create a context, and respond based on that. An example: if you are on vacation, and you search for "something to do", Google Now is able to suggest places to visit or things to do.
How is it able to know that? A good clue would be a change of location. If you are always in a particular location at a particular time on a Thursday, for instance, Google can note the change in the location and know that user intent might include a place to visit or a restaurant to eat.
Google Now can do more than personalizing search. It is also able to perform some actions on behalf of the internet user, like launching apps and playing music. All these functions make it an agent, something that works "on behalf of" the end user. Google Now will work on Android 4.1 and later.
It is also important to note that it actually "talks back" to you. If you ask it to read your day’s schedule or tell you the traffic update, it will do so. Google Now is well integrated into Google’s search, thus able to provide accurate, and diverse information.
Perhaps a more intriguing interactive agent concept is the chatbot. What makes it interesting is the fact that statistics from BI Intelligence showed that by the second quarter of 2015, messaging apps had surpassed the use of social networks.
What does this statistic mean?
Chatbots have a huge potential to become relevant in messaging apps. Let’s consider what we have always known as the online "buying pattern". Users go to a search engine, enter the description they deem appropriate for the product or service they need. They click on a website, and if it is relevant, select the product or service they need, add to their cart and buy.
Chatbot simple use case example:
What chatbots are now doing is taking the entire "buying procedure" into the messenger app. All you need is Facebook’s messenger app. The big win for chatbots is that there will have a very good UI, something that search engines and websites have always wanted to achieve. Better still, relevance will increase, and we can now have the internet user’s attention for more than 7 seconds.
Brands like Louis Vuitton and Levi’s are showing very creative ways of using chatbots. The former launched their chatbot which will advise shoppers, especially during the holiday season, which is definitely quite busy. This bot will be operating via Facebook messenger and is aimed at making the customer experience more personal. The bot will provide information on all the products, as well as suggestions.
Types of chatbots
There are two types: those powered by machine learning/artificial intelligence, and those powered by rules.
The latter is very specific and only responds to specific commands. The former understands language in addition to understanding commands.
How Can Websites Help Search Engines Serve More Relevant Results?
You may have noted that we have barely mentioned websites in this article. Does it mean that they have become completely irrelevant?
Absolutely not. Remember, interactive agents like Google Now are deeply integrated into, and powered by, Google’s search technology.
A logical question would be, if Google Now is going to serve you relevant results — for instance, restaurants around a new area — it must get that information from somewhere. Where would that be?
This means that websites have to do more to ensure that search engines, or better still, agents, can easily retrieve the required information.
We will look at one or two concepts that websites can use:
This refers to HTML added to a web page’s markup using the Schema.org vocabulary. It is a way of adding structured data to web page content that helps search engines parse the meaning from human language. This is done by the connections formed by the linked data in the markup.
It points back to semantics, meaning.
b) Original, high-quality, and relevant content
This content will need to address the user intent. It is therefore important for webmasters to really understand their target audience, and their needs. Using artificial intelligence tools to link data and entities on your WordPress site with WordLift will help build your own knowledge graph for your website, relevant to your site’s topic.
Interactive agents are able to then query this knowledge graph and return more relevant results. Making your site talk back to search engines and give them the answer to the user’s question.
Where can search engines get answer from? Here’s a nice little illustration that WordLift provided in regards to how your text can mean so much more.
All in All
Now that search engines have the ability to "talk back" to humans, search will become more relevant, easy and fun. The interactive agents will be able to do things on behalf of the users, a core principle of the semantic web. Websites, at least for now cannot be left out of this revolution, as they are still the primary sources of information.
If schema markup is added and high-quality, relevant content published, then the semantic web will grow and achieve its purpose - to enable computers and humans to co-operate.