iSA Toolkit

The Intelligent Choice

Grapeshot's SDK provides a suite of functions to improve explicit and implicit search fucntions inside your application - the "intelligence search architecture", or "iSA Toolkit"

In today's Internet 2.0 world search underpins how content finds the user. No longer is search about users typing in just one or two keywords. For example, Google Adsense picks words on pages to trigger related advertisements and Amazon suggests related books as part of the e-commerce transaction. These are the type of intelligent functions that can add value and performance to your own application design. Grapeshot packs in a suite of solutions to maximise the ROI of using implicit search, in clever and intelligent ways.

iSA Toolkit

Grapeshot ranks word, not documents. Grapeshot gives publishers a new tool to understand their audience profile. Advertising needs to achieve high click-through rates to achieve results for both publishers and advertisers. That means getting the right ads to the right users. Grapeshot can create a top profile of best words for each document, and arbitrage requests to ad servers to place the most "in context ads" alongside what the user is reading. Run-of-site category-based advertising can be switched to granular, more personalised, ads based on words the user has read in his or her clickstream (correlating text ads and display banners with context of the page history). Grapeshot can also profile each user, based on the pages he or she reads, and for example, identify the top 10% segment of an audience that like certian things, e.g. "Apple ipods".

iSA Toolkit

Grapeshot ranks words, not documents. It means Grapeshot can pick out the best words in a document, so when a user navigates through a document, the words in the clickstream can be collated around a user ID. As a user reads more content, the WordRank weights for each word constantly change, adapting to that one unique user. The user's unique profile is used to steer fresh content to the one user - transforming the ability to target better ads, push a personal set of headlines, or skew the ranking or priority of content delivery to people, as individuals.

iSA Toolkit

Grapeshot uses WordRank to identify the best words in a document, and this technique can also be used to select the most significant sentences. It means a whole set of pages can be shrunk down to a few best sentences - a great way to skim read documents or create an abstract.

Grapeshot ranks words, not documents. The WordRank word weights can actually be personal to each user, based on the user's past clickstream interactions (especially in personalisation systems). This means summarisation can also be personal, so that one document can actually be summarised in two different ways for two different users - with Grapeshot sensing which are the more significant sentences for each user!

iSA Toolkit

Categorising documents can be onerous on human time, so automatic methods from Grapeshot dramatically improve productivity.

Grapeshot ranks words, not documents. It means Grapeshot can treat a whole document as a query, and match it against a dataset of categories, defined as a set of sample pages, each with words and themes. A new document is matched against a set of categories, providing a range of probability confidences for suitable categories for that one document. This Grapeshot process can categorise over 200 documents per second, and be used against several different taxonomies simultaneously.

iSA Toolkit

Organising information into collections of similarly themed documents can aid rapid assimilation of large document investigations.

Grapeshot ranks words, not documents. It means when a search result is returned, Grapeshot can do more than just list results. Firstly, Grapeshot knows the best words in each sentence of each document - so Grapeshot can be used to pluck out the best few sentences as the caption. Using AJAX methods, the user organise the search results into groups. Instead of relying on any predetermined fixed thesaurus or categorisation system to fix the organization of grouped documents (into categories), the user can have full control on manipulating the degrees of similarity between documents, quickly establishing how different documents overlap in varying degrees of concept. Grapeshot uses a top of WordRank weights for words in each document, and can readjust weights and display of conceptual overlap interactively as the user manipulates the interface in real-time.

Better by design

Grapeshot is different from most search platforms. Built by Dr. Martin Porter, "the inventor of robust scalable stemmer algorithms that underpin search", Grapeshot doesn't imitate the competition, but innovatively reinvents the field. Porter has taken a lifetime of experience and started from a clean canvas to build the optimal architecture for search, instead of piling unwieldy work-arounds on older methods. This allows several unique benefits:

• Lighter weight - Grapeshot has everything you need and nothing you don't. Avoiding the bloat of other platform "solutions", you can develop more efficient approaches to delivery of your search enabled application.
• Written for OEM applications - Grapeshot has an extremely flexible, extremely well documented, extremely simple API. You can integrate faster and more seamlessly than with competitor's more complex, after thought adaptations to OEM needs.
• Performance and scaleability - Grapeshot's founders have a history of setting the industry benchmark for scale. The elegance of the indexing and retrieval algorithms allows Grapeshot to manage far more documents, at far greater speeds than any other competitor.

Conclusions

Grapeshot's intelligent search architecture delivers unique value to application developers companies: improved performance (better search for your customers), improved asset productivity (quicker to learn and integrate for your fast development cycles), and improved operating productivity (easier to support.) All of this adds up to real competitive advantage for your business.