Cover photo

Using AI for personalized search

Using AI for personalized search in Web2 and 3

Using AI for personalized search

I. Introduction

AI is rapidly revolutionizing the search process, allowing for more personalized and tailored search results than ever before. AI-powered personalized search algorithms can analyze user behavior in order to identify factors that may influence the value of different search results and deliver more contextually relevant and meaningful results. These algorithms can also help to filter out irrelevant information and ensure that the most relevant and accurate results are delivered to the user. However in Web2, these algorithms are controlled by large corporations whose ultimate goal is to present tailored adds and sell you more products on behalf of their customers (the advertisers). In this article, we will explore how AI can be used to improve the search process, and deliver more personalized results, both in the context of web2, and also web3.

II. Overview of AI and its use in personalized search

The core principle of AI-powered personalized search is to analyze data sets (big data) in order to identify patterns and trends that influence the relevance and accuracy of search results. The algorithms can take into account user behavior based on factors such as location, search history, and preferences. By doing so, AI-powered search algorithms can deliver results that are tailored to the user’s individual preferences. In the case of Web3, we can go even further by looking at user social graphs (posts, comments, likes etc.) as well as monetary transactions. In addition, the algorithms are not biased by having to present personalized ads to sell products.

III. Understanding User Social Graphs

A. What is a user social graph?

A users’s social graph is made up of interactions that a user has on social media, such as posts, replies, likes, comments, tips etc. In the case of web2 (twitter, facebook etc.) much of this history is public and searchable, but not usually available thru API’s since the platform owners want you to go thru their interface so it is generally harder, or more expensive to get. In the case of web3, the private data can be controlled by the users. There is also lots of public data which includes the transactional history publicly available on the blockchain. This can be valuable in determining behavior such as spending patterns.

B. How AI can be used in conjunction with user social graphs

It seems to me that all this information can be useful in tailoring personalized search results. Opensource algorithms can analyze the data in order to identify user preferences and interests, based on their social graphs, allowing for more relevant search results. AI-powered algorithms can analyze current trends in topics that may be related to the user’s behavior, ensuring that the most up-to-date and useful results are provided. This can help to eliminate irrelevant and outdated information.

Here is a list of Web3 projects that are working on personal social graphs. This is taken from the article: Getting ready for the Web 3.0 Social trend (on Mirror)

“Social graphs”

“The social graph visually resembles a spider's web and stores the chain of connections between different participants. These connections are a major social media asset monetized by legacy social companies through targeting, advertising and selling data to third parties.

Existing monopolies are not interested in open and free-flowing data because they get the most profit out of a closed and controlled system. However, this makes them ineffective, rigid and leads to stagnation and inevitable collapse of user bases and revenue.”

Here is a list of some web3 social graph projects, taken from the above article

Social identity apps can verify that the user accounts are valid. I am currently using, and my social accounts are linked here:

III. Examples of AI-Powered Personalized Search - I have been testing out this new search engine, and it looks very interesting. Not only can it include social information in the search results, but it has also combined AI tools like Chatgpt, AI Image generation from text, a writing tool to assist with articles and blog posts for example, and other tools for developers. You have more control over the sources that are searched, and you can chose from 150+ integrated apps.

They also provide a way for developers to modify the UI and produce custom searches.

In the case of, the social media data is based on Web2 content, but would be even more powerful if it could include web3 social graphs, social feeds thru open API’s and monetary data as well.

IV. Conclusion

Here is a good article which talks about the current web3 search tools: - Farcaster social graph search - Mirror search (listed in the Cyberconnet social graph ecosystem) - search in Lens, Farcaster, Mirror - Wallet search - Lens protocol search - Mirror search - DeSo search - Dune search

Web3 searching tools by using scrolling feeds: - excellent functionality and convenient to use on a computer - nice phone app - social wallet with good UX\UI, recently launched a mobile app

Overall, AI-powered personalized search is a very promising technology, and I believe it will become an invaluable tool in the future of web3. It has the potential to deliver more relevant results and improve the user search experience. However, it is also important to remember that this technology can be abused and used to manipulate search results so that is why it needs to be opensource and transparent . So far, I have found to be a valuable tool, and I think that it may even has a chance to challenge google search. There is more of a learning curve, but it is more flexible, customize-able and gives you more control over your searches. It will be interesting to see how it plays out.