Data Fusion is the art of intelligently linking data from disparate data sources to create a holistic picture that serves a business scenario.

For example, a financial analyst might want to trawl the latest analyst reports for the companies in his portfolio and combine them with the companies’ latest SEC K10 filings. A product manager might wish to track customer reviews of his latest product across a range of sellers. An app developer showing real-time movie ticket availability might want to include plot summaries of movies from IMDB and reviews from Rotten Tomatoes along with movie showtimes. A healthcare analytics company might need patient information merged and reconciled across multiple health data repositories.

The data sources often don’t have to be external. It’s not uncommon for data to be in isolated silos within an enterprise. Obtaining a complete picture of their customers by linking store transactions with online browsing is a big pain point in many e-commerce companies.

While the need for data fusion is clear, the process of fusing data today is far from ideal. In many cases, it’s manual. This is, of course, error-prone and brittle. Even if the process isn’t manual, it’s typically done in an ad-hoc manner that doesn’t scale.

At RedSieve, we have extensive experience working on Data Fusion solutions across a broad set of domains - e-commerce, financial data, reputation management, identity information, health care - to name a few. Therefore, we have built the general-purpose RedSieve Data Fusion Platform from the ground up. The platform is highly scalable, effortlessly able to handle millions of data elements in realtime. The platform is easily configurable - with appropriate configurations, it can be made to work across a series of domains and use cases.

Example 1:

A recruiting company needs to construct a complete profile of potential candidates by linking data about candidates from multiple social media.

Example 2:

A online comparison shopping app would like to consolidate data about a product - prices from multiple vendors, professional reviews, specs from manufacturer sites and comments from forums.

Example 3:

An e-commerce portal with both a physical and online presence would like to learn more about their customers by merging their physical store transactions along with their digital ones.

Facet Extraction and Data Enhancement:

A great shopping experience begins with a great product catalog - a catalog with clean, authoritative, structured, organized information.

The RedSieve Facet Extraction Engine uses proprietary technology and algorithms to extract structured, standardized information from raw unstructured blobs of text. The engine can automatically detect synonymies - thus “blk” is the same as “black” and “14 oz” and 14.0 ounces” are the same. The engine scours the entire catalog and automatically creates and populates new attributes for each product.

With such structured facets, the search can be a lot smarter. When a customer searches for a “Vintage wooden espresso writing desk”, the most relevant product can be surfaced. Moreover a lot of these attributes can be used for “faceted searching” on the left navigation panel when search results are displayed. And more powerful recommendation algorithms that can leverage rich faceted data can be deployed.

Autocomplete done right:

A good search autocomplete solution will ensure three things: 1) A context is included along with the suggested search query - “Flower pots in Home Furnishings”, “Harry Potter in Instant Video” instead of just “Flower Pots” or “Harry Potter”. 2) Redundant suggestions will not be shown - there’s no point in suggesting “Flower Pot” and “Flower Pots”, a typical characteristic of naive solutions. 3) Using a suggestion always leads to good search results. The RedSieve Autocomplete Engine analyzes query patterns and results and provides a highly relevant and performant auto-complete feature for search.

Competitor Data Feeds:

How are you faring in comparison with your competitors? How are their prices? How rich is their product catalog? How does their assortment and inventory compare with yours? Are you carrying all variations of a given product? Are you carrying a comparable number of products in a given category?

Identification of Misclassified Products:

Are knitting needles egregiously classified under Shower Curtains and Accessories? Would a customer browsing under “Stylus and Pens” stumble upon a portable golf training aid? Are all cameras in the cameras category actually cameras? Or did a few camera tripods accidentally creep in? Our solution scours the product catalog and provides a recurring report on misclassified products.

Third Party Merchants Marketplace Integration Pipeline:

A full-fledged integration pipeline that incorporates and integrates third party merchant feeds as you work on expanding your selection and bring partners onto your platform.

An event-based metrics platform:

What are your customers mostly searching for? How many search queries lead to no results? How frequently does it happen? Is there anything common to those queries? How does a particular search query’s performance track over time? Over last one week? Last six months? Did it’s performance suddenly degrade in the last couple of days? (Perhaps someone accidentally changed the image of one of the query’s best selling products to something offensive).