Enterprise Semantic Search

 

Client Overview

The client was an IT company that specializes in outsourcing IT support services in America. This company provides solutions to computer hardware and software issues. They offer phone, online and real life support to their customers.

 

The Problem

Our client opened more offices across the Florida state in a span of six months. Our client did not have enough time or resource to provide training to new employees in a short of amount of time. IT Company’s branch managers decided to create a standardized online training/evaluation section on the Enterprise portal. The branch managers even encouraged the more experienced employees to voluntarily post documents their advance IT knowledge in the portal. Over the next few months, the company’s knowledge base was filled with improperly tagged/titled free flowing articles.

 

“There were so [many] articles and the knowledge base was a mess. It was difficult to find the information [that I] need.” - said a new hire.

 

“It was frustrating to go through [all] the documents [in the Enterprise Content Manager (ECM)].” – Another employee complained.

 

The Solution

The owner found out about Techome China through Linkedin, and he requested Techome team to revolutionize the Enterprise Portal. The client and Techome agreed on a fix price short term project. The project goal was using Web 3.0 and web 2.0 technologies to implement a semantic enterprise search. In order to efficiently complete the project, a team of technicians were formed from both sides.

 

- IT Company’s Technicians

1. Will survey their employees to come up with a more organized structure for the knowledge base and evolution section.

 

2. Filling proper tag, Microformats, RDFa (Resource Description Framework), and Meta information for the free flowing articles.

 

- Techome Team

1. Creating a web application to add Microformats, RDFa, and Meta Tag into the free flowing articles.

 

2. Using entities and relations to create a Semantic Search Engine.

 

- Semantic Search Engine

The ECS already have a basic Search Engine. The Search Engine crawls through the entire ECS only returning results based on the Meta header - and not all articles and documents are properly labeled. For example, some Meta headers were just duplicates of old articles and others were left blank. Techome Team requested the IT technicians to look through mostly used articles and filling the correct headers. Then Techome team implemented Microformats such hCard, hCaldender, rel-tag, votelink, and hReview. For example, using hCard, the Search Engine could easily pick up article author, editor, information and so on. With article topics and styles that were not covered by Microformats, our engineers marked them with RDFa. RDFa provides xHTML attributes which creates machine readable hints. The addition of RDFa allows the Search Engine to return relative and reference topics that search keyword does not cover. Aside from Meta, Microformats, and RDFa, our web developers also implemented like and dislike function so that the employees can rank the usefulness of the articles and further improve on the quality of the Semantic Search.

 

Conclusion

Using Meta HTML headers from web 2.0, web 1.0, the technicians were able to sort some of the articles accordingly. Through Web 3.0’s Microformats, RDFa, and the like and dislike functions, the Semantic Search were able to return higher quality and related articles.

 

“Thanks to Techome China, we are able to work faster and our new hires are better prepared for their job.” - Said a branch manager.

 

Read More