What are the requirements of the next generation of search engines?
3. The BingoBo idea: SENG (Search Engine: Next Generation)
While we are shooting in the dark for new solutions, Web 3.0 sheds light on the direction of the next generation of search engines. After analyzing the existing problems and the solutions shown in previous parts of this topic, I found that future search engines will have some major characteristics, which are listed below, to provide more relevant search results tailored to user needs.
- Empowered with human intelligence
- Pull vs. Push
Designing a system with the above merits in mind and put them together to work as a more efficient and feasible solution towards Web 3.0 is a big challenge.
From the very beginning, the system is required to have a hierarchical structure with multiple layers of categories because human intelligence involves this type of structure in order to memorize more information in a limited brain space. This requirement came out before Yahoo Directories was introduced in its portal web site.
Secondly, the system must be distributed as information all over the web is tremendous and would be hard to scale within a centralized system. The Agent Technology developed by the DARPA's Knowledge Sharing Initiative made the multi-agent system architecture ready to work for this kind of distributed collaborative peer-to-peer communication between autonomous agents and brokers.
Third, the distributed agents must be geographically aware so that each agent can collect the local information from the particular local area and allow users to roam around from site to site. This was conforming to the Web 3.0 concept and was created before Local.com made its geocoded and geocodable algorithm to collect address information from the web pages, which has the same local concept but under a different mechanism. It also resolved the critical security issues of the mobile agent. Instead of making agent roam across different locations, it allows agents to sit in different locations and let the users roam from site to site. BingoBo's distributed local agents will help businesses to integrate their product and service data with the BingoBo system to have automatic updates.
Fourth, being personalized is a mandatory feature for Web 3.0. Therefore, BingoBo created its unique way to realize the self-organized personal storage requirement via a pair of public and private views, rather than the browser bookmarks or online bookmarks that can hardly help us to store the useful information under a consistent data structure. In addition to saving search results, users can save their personal information as well, such as their family tree, medical history and contact information.
Fifth, online privacy has become a more serious issue. The system will give better user experience if search engines respect user privacy and do not track their online behavior. It will also benefit the users in the long run to adopt "pull" technologies to replace "push" technologies and give users the privilege to pull the relevant information according to their needs.
Sixth, online category starts from the Usenet, which provide the readers with classified news and other types of information. Unfortunately, Yahoo Directory, Craigslist and other classified websites didn't move further to help users save and share their favorite information in an efficient way. Thus, they are unable to serve as a common platform for users to search relevant information reliably. Moreover, it's hard to establish a standard product catalog definition across various industries. Previous efforts, such as Rosetta-net, failed to be widely adopted. Actually, such standards may not be necessary. Like the product SKU system, nobody can guarantee a unique SKU to be issued, but it still works well with trillions of products on the market. So we will build a de facto product/service master along with the user feedback.
Last but not least, the system must provide quality search results to save users time and money. It will help users find discount information easily and filter information based on certain criteria. It must have a way to show the search results reliability that are "voted" on by the users instead of machine algorithms. BingoBo provides a unique quality indicator by the number of subscriptions in addition to the number of hits. It shows how many users seriously think the piece of information with links is important and useful to be saved into their Private Web Plus, it allows users to share their favorite links to the public and friends. Over time, the human empowered part of information will dominate the well-organized public web.
Based on the above requirements, I worked out an overall design to knit the pieces together with features supporting the most common definitions of Web 3.0. In the following chapters, I will discuss how the requirements have been translated into technical functionalities and implementations.