IBM's open-source OmniFind search engine easily turns intranets into a searchable format for employees and customers.
One of the biggest problems that I faced recently was trying to come up with a good solution to create an indexed search database across multiple intranet formats. Along with static HTML pages, we use WordPress to host intranet news and some content management, wikis are used to hold frequently changing information that users can edit themselves, and network shares are utilized in some cases. Although each of these technologies often presents a tool to search their own implementation, I wanted something to index all of them with one interface to search from. Otherwise, a user has to know which Web site or format the material is in and then go search using each individual search feature.
A few open-source projects out there offer indexing. After doing some research and testing for a few weeks, I found that most of the open-source projects involved too much work to set up, configure, and use. Very complex configurations, a lot of manual effort, and lot of downloading and compiling plug-ins just didn't seem to be a greatly efficient way to tackle this problem. Finally, I stumbled upon IBM's free OmniFind touting a "3 click setup" with a working configuration in minutes. I was pleasantly surprised by what I found.
Prerequisites
IBM's free version supports indexing of up to 500,000 documents and has five different organizational units called "collections." A paid enterprise version is available if you need to index more documents. If interested in the commercial version, you'll find resources on the link below to contact IBM. Supported operating systems are RHEL 5, SUSE Linux Enterprise 10, Windows XP, and Windows Server 2003.
IBM recommends at least a 1.5GHz processor, 1Gb of memory, and 80Gb of disk space to index up to 20,000 documents. Depending on the age of the hardware, you may also want to consider a dedicated server rather than a shared host, since indexing can be resource-intensive at times. Downloading IBM OmniFind requires only that you enter a valid email address and answer a few general questions. The download page also provides installation, administration, and developer guides.
Web Interface
The Web interface is about as simple as you can get. This makes configuration and setup extremely easy. The very first page to display after you log in is the Collection Status configuration page. As mentioned, you're allowed up to five different collections. Each collection can contain Web sites and file directories to index.
Figure 1: The collection configuration page looks like this. (Click images to enlarge.)
To enter Web sites to be indexed, simply type in the URL of the site and click the Add Web Site button. File directories can be added through the local file system by browsing to the folders you want OmniFind to also attempt to index.
Cache Settings
Under each collection, you can choose whether or not you want to store a cached copy of all documents that the index creates. The positive in enabling this option is that it gives you the ability to retrieve documents even if the Web site or folders are offline. If you decide to enable this option, make sure you have adequate disk space on your system, though, since it creates a local version of the files. Disabling the cache option speeds up indexing since it doesn't have to create the extra files necessary. When the cache option is disabled, you will not have the ability to see a cached or HTML version of the document under the search results.
Metadata and Synonyms
The metadata screen allows you to edit default fields or create new ones. This nice feature allows you to create custom terms for users to search on. Metadata types that you can choose from are text, decimal, and date fields. You can then map a metadata name to a field, giving users additional search capabilities.
Custom synonyms give you control to expand similar queries to match other words. As the documentation explains, you can input the terms "films," "movies," and "moving pictures" as synonyms. If a user searches on "films," the search will expand to look at the terms "movies" and "moving pictures" as well, dramatically increasing the chances of a successful search result being returned.
Customize Search Page
The default search page contains both the IBM and Yahoo logos. Under the page layout section, you can easily upload your own company logo to replace it, though. The fonts, colors, and layout of the search results can also be changed. You can remove certain features—such as summary, URL links, or the cached results—from displaying at all. Click around the page as everything is rather self-explanatory.
Figure 2: Customize how the search page is displayed.
Performing Searches
The default search page can be found at the URL http://yourhostname:8080/search. The search page has a similar look and feel to any typical search engine you visit on the Web, such as Google or Yahoo. There's both an advanced search and a preferences link. Since the free version is heavily sponsored by Yahoo, you'll find lots of links along with copyright and trademark information dealing with Yahoo.
Other than that, you can search away on whatever you've configured OmniFind to index. Given this utility is not only free but also easy to install and configure, it's worth a try for your indexing needs. With all the more time and effort it took to set up, I've been very happy with the ability to have one location to search many content management systems.
LATEST COMMENTS
MC Press Online