SRCH2 is a new search software that is specifically designed to support fast, instant search, which can bring “Google Instant” experiences to your applications.
Frequently Asked Questions:
What does SRCH2 do?
What are SRCH2's feature advantages?
SRCH2 offers a compressed full text forward index in memory. This results in super fast, simple, error tolerant, highly configurable forward recommendations. It is built in C++. It is algorithmically advanced, enabling configurable error correction based on edit distance, and rapid geo search based on stored geo data at the index nodes.
What can SRCH2 do that I can't get out of Lucene? Lucid? Elastic Search?
SRCH2 is specialized in type-ahead search. Compared to the solutions that support such a feature by doing wildcard queries, we provide more control on the ranking even in the presence of prefix conditions, possibly with typos. Our unique in-memory techniques enable higher performance and richer features. Furthermore, we also support built-in access control and smart ranking learning.
What's the advantage of a type-forward index in memory?
It’s fast. If you were to try doing it with Lucene, you would see a slower search engine. Add high volume of queries, and it’s a big problem. We configured Lucene 4.1 (released in Jan, 2013) to do forward search, with geo, and fuzzy error correction, and found that SRCH2 was 31x faster on an identical stack, with a 5mm record db.
How else is it better than other search softwares?
Relevance and customizable ranking specifically designed for instant search and error correction. Configurable forward recommendations. So a retailer with hundreds of locations can display only results available locally. Try doing that with your search software.
How hard is it to deploy SRCH2?
Search has gotten a bad rap. It’s complex, expensive, time-consuming. We get it. But if the complexity is kept to a minimum, and if the uses are specific, and if the data is clean and accessible, then getting started is a snap. Take a typical e-commerce site. The data exists already, in a product database. It’s already accessible by the existing search tool, whether it be open source or other. As long as that’s the case, launching SRCH2 takes a few hours, if that. Just download the code, index your data in JSON, configure search your way, and run. We have a full set of step-by-step instructions, with pretty pictures, here. How can you resist?
Does SRCH2 help integrate its search software?
Yes. All customization is documented, and we are also available to help you configure and launch. SRCH2 also works with trained and certified integration partners to help you with custom or UI work as needed.
How does SRCH2 work with SQL?
SRCH2 has built connectors to directly index SQL databases.
Does SRCH2 work with MongoDB?
Yes, very well. Like most search engines, SRCH2 indexes your data in JSON format. You just dump your data into JSON, and then run SRCH2 to index it. Then configure and launch. This process is identical to Lucene, Elastic Search, Lucid, and Endeca.
Are free trials available?
Yes. Click the “TRY SRCH2” link above and just download it! (Seriously, what are you waiting for?)
How do I configure and launch SRCH2?
We offer comprehensive documentation. You can see it by clicking on “DOCUMENTATION” below.
How can I use Customizable Rankings to solve business problems?
Our clients are doing many things with this feature. Want to decide what to show to site visitors at your e-commerce site? Based on conversion rate? Profitability? How about inventory on hand? How about LOCAL inventory on hand? Yes, we can do that. If you can put the info in your database, SRCH2 can use it to adjust type forward search results.
Does SRCH2 offer a RESTful API?
What are the ways SRCH2 can be configured?
SRCH2 is sold as an enterprise license, served by you. Or it can be deployed in a platform setting. For example, it can reside on handsets, interacting with a server. Or set-top boxes interacting with a central server. Finally, SRCH2 is also available as a hosted search tool.
Why can't I just use Google?
Google Commerce Search charges for the number of search queries, charges for not showing Google ads, charges for configuration and customization. Many larger enterprises have firewall concerns, making Google an impractical choice. Finally, the big advantages of type-forward search are not there, because Google’s type-forward is based on query logs, not db indexes.
How is your type-forward different than Google's?
Google is great. Many of our team and investors are from Google. But their search software is designed to work on Google’s servers, using their unique data sets. They do type forward using query logs, tied to massive server farms the size of Vermont. It is awesome. And it’s not a resource you have. SRCH2 does instant search using a cached forward index of your data. The results are faster than anyone else, and more relevant because they show your actual data, not other people’s queries.
I see type-ahead using other search engines. Why is SRCH2 better?
Search boxes can be configured to offer synonyms for common queries, and even misspellings. These may offer suggestions, but unfortunately, the hit-or-miss nature of the solution leads to poor relevance. SRCH2’s full text index offers much more relevant suggestions, because it’s the real output of a powerful search engine, spitting out results based on a full index of your database. We’re not guessing at what users might type. Our type-ahead is the actual output of our search engine. That’s what lets it do not just type-ahead, but much more. Error correction, customizable rankings, geo suggestions, and more. Once you see what a cached forward index delivers, you will start seeing the problems with search boxes everywhere.
Can error tolerance be configured?
What are the trade-offs when configuring?
As with every advanced search feature, there are performance trade-offs. Features can be turned off and mixed to meet your specific needs. Configure the search engine the way you want your search to appear.
Why is SRCH2 better for mobile?
Searches on mobile devices are harder because your fingers are not on a full keyboard. You make mistakes, and each character entered is a pain. Instant forward search lets you type less, and error tolerance fixes your typos. In addition, many mobile searches involve geo targeting, which is a well known pain point. Finally, mobile sites and apps also have to deal with network latency and memory limitations. All these things slow you down. SRCH2 was built to speed you up. We address every one of these issues, with a super-fast search engine. Did we mention it’s kinda fast?
Is there a standard definition of Instant Search?
What we think of as “Instant” type-forward search only works if it is faster than the human doing the typing. Researchers typically peg this at 100 milliseconds. Meaning the results have to be shown in under 1/10 of a second in order to have any value. Slower than that, and you slow the user down. Plus you choke your server with a new query with every keystroke, so there is a big cost of slow “Instant” search.
Does SRCH2 provide any additional benefits?
Yes, cost. If you have many visitors, there is a substantial server cost you bear when you deploy “Instant” search. You paid to develop it. It cost time and money. You weren’t building other great features while you built it, so there was an opportunity cost. Now, on top of everything else, here’s something you may not have considered. In your “Instant” search, every keystroke creates a new search query, requiring more processing. Add a lot of records, or a lot of visitors doing a lot of searches, and this server load spikes. Suddenly, you’re hitting your server capacities, and have to build in new failovers and spike tolerance. This is costly. Likely all of these costs are a lot more than just deploying SRCH2. We’ll give you something that’s better, faster, and cheaper. Believe.