What Is Neural Matching? Everything You Ever Wanted To Know!

AI is here and Google cannot but use it in its search engine. Google announced a Broad Core Update to its algorithm calling it Neural Matching. The announcement was made on 12th March 2018.

Perhaps the best way to understand this Broad Core Google update is to take a cue from its recently published research paper. In this paper talks about matching search queries with web pages using only the queries and the web pages.

In keeping with the tradition of discussing the latest algorithm, Danny Sullivan of Google elaborated in a tweet:

The focus is to show how AI can be used to understand different meanings and catch the varying nuances of the same word to connect them better to related concepts.

What is Deep Neural Network?

Deep Neural Network is a technology that simulated activities of the human brain – precisely the different recognition patterns and the way inputs pass through different layers of activated neural connections.

Deep Neural Network then has an input layer and an output layer with one or more layers hidden in between them. These hidden layers carry out the work of sorting and ordering that is often referred to as “feature hierarchy.” This technology is used mainly when working with unstructured data.

Deep Neural Network is often synonymous used with Deep Learning where machine engages in the deep learning process using AI to gather, classify and order information.

Google’s Neural Matching and its connection with Deep Neural Network

Google is working on using the algorithms of deep neural network method in its Neural Matching. In its research paper Deep Relevance Ranking Using Enhanced Document-Query Interactions, it talks about Document Relevance Ranking and terms it as the Ad-hoc Retrieval. It is explained as ranking documents from a large collection using the query and the text of each document only. This contrasts with standard information retrieval (IR) systems that rely on text-based signals in conjunction with network structures.

The new algorithm in Neural Matching aims at matching search queries to web pages using only two factors: search queries and webpages. As this matching is all about relevance matching, they do not have to be promoted to the top positions using keywords or links.

This new algorithm is all set to re-rank pages that already appear on the search engine pages. Google makes it amply clear with its tweets:

What does “building great content” mean?

The relevance factor brings in the use of synonyms as Sullivan already explained with his tweets. But does it mean the random use of synonyms and spamming a page with the keyword variations? Certainly not. This is amply clear from his tweet:

Google wants to understand synonyms as per the context, subject and the meaning of the content in a webpage. Clear and consistent web content is more important than using mere synonyms of keywords. Because Google has officially declared that it is going beyond mere keywords and their synonyms to find the relevance of a page related to the queries of a visitor.

The excerpt below taken from Google’s official blog site explains it well:

Summary: Google’s Broad Core Update Neural Matching uses AI in ranking pages not using traditional methods of using keywords and links. It tries to understand the concepts that lie beneath and match them to the webpages carrying similar concepts.

If you are wondering about the changes that have to be made, then pay more attention to user intent, the type of queries that your target customers are likely to throw up and formulate relevant content accordingly.

Leave a Reply

Your email address will not be published. Required fields are marked *