When it comes to business process automation, it’s not a big problem to code the workflow; as business rules can be formulated, and thereby turned into code. But you have to do much more when you need to apply these rules to your business data. In search of the perfect solution DataArt checked the progress with a resource called WolframAlpha that is gaining in popularity.
Every working business has a certain amount of good quality business data (read business knowledge) that has been typed in manually, crawled, grabbed, bought, or accumulated other way over the years of a business’ existence. The question is – where to get new data?
Though the Internet contains a giant amount of information, it is still hard to obtain it automatically, because data is often unstructured, unreliable, hard to retrieve, and poorly or not even tagged at all. Moreover, data is typically drawn from different web sources with a need to develop your own data extractor (data crawler).
The problem is much more acute for data-related businesses than for those who just handle the data. For instance, a crowdsource-driven social network resource may not suffer from poor data quality at all; whereas a system that collects and analyzes the pricing info of a customer market would be extremely vulnerable to data purity.
Until recently every system that required high-quality data at the input stage had to have its own means for enabling it. These means, besides multiple properly built data extractors, typically include multi-step automatic and even manual data validation, purification, and approval modules.
Needless to say that these modules that are present in every similar system take substantial time to develop and debug, they increase time-to-market and cost of making a new release, along with the necessity of having the required personnel for manual approval and finalization of data acquisition.
Obviously, a data vault with a unified globally-accessible data fetching API could be a solution. It should have a big amount of data, and a quality that would fit the building of the most demandable business-processes. This would never, definitely, be free. But having a moderate subscription cost for such a third-party service looks much better than building and maintaining your own copycat system.
Such a system now exists, thanks to the appearance of the Internet-based knowledge store Wolfram Alpha.
WolframAlpha is a computational knowledge engine developed by Wolfram Research and launched in 2009. It’s a powerful service that responds standard user queries with more than 10 trillion sources of continuous updatable data.
The sources curated by WolframAlpha belong to completely different branches of human activities – technologies, science, and industries such as Mathematics, Astronomy, Geography, Finance, Nutrition, Medicine, IT, etc. The computational power of WolframAlpha is comprised of more than 50,000 types of algorithms & equations to make query relevant, clear, and finished. In spite of the huge variety of output data types, the results are supplied in a user convenient view by means of grids, plots, images, and interactive cdf blocks.
To make everything clear, CDF (Computable Document Format) is an electronic document format created by Wolfram Research. Its main goal is to refresh content by updating it in response to GUI interactions. It means that users can scale plots or turn volumetric figures just by means of cdf-block inner elements such as sliders, menus, and buttons. And all the computations are client-side! Everything looks clear and simple – the developers of this product made a well-thought out interface.
Surely, it’s the best data engine I’ve seen.
Interacting with WolframAlpha features via its APIs is a great feature for applications to take advantage of. They can simply access all of the available data and have it computed and visualized according to a user’s requirements.
Nevertheless, access to such a wide variety of information means there is a price to pay. The running of a human-language query, and computing and structuring the results takes some time, and delays can be quite long and the total value of information obtained too high.
DataArt has great experience of development with the integration of external sources. Our engineers already know how to build queries for the WolframAlpha API and handle the results correctly to increase speed, and to program a search function to filter the unnecessary information from that which is important. The connections with the Alpha team provide good technical support and reduce feedback delay.
The WolframAlpha API was successfully integrated with DataArt’s R&D food recognition project we’ll write about later.
As a store of large amounts of information, WolframAlpha opens up new horizons for business projects. Its implementation as part of a desktop or mobile application will provide more accurate results for data requests when used. Our experience has shown that this service has the ability for in-depth analysis and provides a maximum of accurate information when used in all sorts of projects that require deep global search functionality.