Immobilien Scout 24 Intelligent automatic data analysis for millions of documents
German real estate portal improves its service offerings with InterSystems technology
ImmobilienScout24 is by far the largest German language real estate Internet portal, with more than 10 million users each month (as of January 2013). The Berlin-based company has about 600 employees and is part of the Scout24 Group, a subsidiary of Deutsche Telekom. ImmobilienScout24 plans to digitalize apartment and house hunting. Right now approximately 45,000 commercial vendors and about 73,000 private persons take advantage of the portal to present their real estate offerings – with detailed descriptions, photos, and floor plans. Each month, more than 1.5 million different apartments and homes are listed on the portal and these listings are viewed more than 300 million times. Furthermore, owners and interested parties find a wide spectrum of real estate information and services, including support for questions about financing, refurbishment, and renovation as well as moving services.
Online real estate appraisal
An important instrument for winning and retaining new customers both sellers and potential buyers is the innovative online real estate appraisal. Potential buyers as well as interested sellers or real estate agents can use this tool to find out the current market value of selected properties and get an overview of market prices in a particular region. This type of online value appraisal is considerably less expensive than a professional assessment, which usually combines several valuation methods.
- Quality of data
- Unstructured data
- Analysis of free text
- Big data
The online appraisal uses what is known as the sales comparative approach: initially a minimum of ten objects from the ImmobilienScout24 real estate property database are identified that are comparable to the property to be appraised. These objects must be of a similar type (i.e., single-family home) and be located in a comparable area.
Differences in fixtures and fittings are taken into account by applying specific surcharges and discounts respectively. The corrected values of the comparison properties are then averaged to find a mean value, which allows for a dependable statement about the market value of the property being appraised.
Database of 180 million listings
The reliability of the value appraisal naturally depends upon the available number of suitable comparison properties and the existence of sufficient information about said properties. This information includes detailed data about the location, year of construction, year of renovation, condition of the property, fixtures and furnishings, heating system, energy efficiency, parking availability, and whether the property is accessible for the elderly or disabled. The ImmobilienScout24 real estate database currently has more than 180 million listings. In many of these listings the necessary information actually is included as unstructured text within the listing. Unstructured data is not readily available for automated data processing. The ImmobilienScout24 appraisal algorithm requires clearly classified data structured data for input. In the past, it was often impossible to use those listings with too few structured elements for value appraisals. As a result, properties were often not able to have their value appraised because of a lack of appropriate comparison objects.
InterSystems’s application partner n3 (or “n cubed”) developed a solution for this an innovative application that extracts and structures the required information from the written text in the various listings. Founded in 1998 (the same year as ImmobilienScout24) n3 , based in Oelsnitz in Saxony, offers services and consulting at the interface of data analysis and software development for its clients in the health care industry, manufacturing, telecommunication, research and development, financial services, and other industries. With VisualCockpit, n3 offers a high-performance business intelligence tool based on the InterSystems Caché® high-performance database and the InterSystems Ensemble® integration and development platform. The data analysis professionals at n3 also used InterSystems technologies for their ImmobilienScout24 solutions in this case, particularly the newer versions of Caché with iKnow text analysis technology.
Automated data analysis
In contrast to many other semantic analysis technologies, iKnow automatically recognizes potentially meaningful patterns and relationships within unstructured data without being dependent upon input from the user or specific search terms. Using smart indexing, iKnow analyzes and transforms unstructured text into a network of relationships and concepts. It recognizes what belongs together, is important, or is characteristic within longer, unstructured texts or within collections of documents. The results of this smart indexing procedure are then connected to listing-specific information via smart matching. Finally, depending on the targeted objective, either specific analysis methods are used or rules of use are applied to the data (smart interpretation).
That makes our services even more attractive, and allows us to offer even more customers accurate appraisals for their properties.
Sven Feuerschtte, Vice President Valuation Services
“iKnow provides solutions for a variety of challenges: You can identify the interesting elements within a pool of documents, automatically process information based on similarity criteria, clearly summarize longer texts, and much more,” explains Matthes Nagel, Chief Technology Officer with n3 . “For property appraisals on ImmobilienScout24, iKnow takes the knowledge that forms the core competencies of an appraiser and assigns the information from the listing texts to the appropriate appraisal slots. iKnow makes possible retrieval and use of more than 90 percent of the unstructured data in the ImmobilienScout24 database, and transforms previously unusable data into quantities that the appraisal algorithm can access and use. In this way ImmobilienScout24 improved both the quantity and the quality of the data underlying its appraisals.
The challenge of big data
“With the assistance of the n3 solution, which completes the missing structured elements in the listings created by property owners, we can now call up even more comparison objects from our extensive database for online value appraisal,” said Sven Feuerschtte, Vice President Valuation Services with ImmobilienScout24. “That makes our services even more attractive, and allows us to offer even more customers accurate appraisals for their properties.”
The solution from n3 shows how one of the biggest challenges related to big data can be mastered: the use of all information that still lies fallow within companies’ and organizations’ storage systems because they are located in a variety of MS-Office documents, graphics, emails, and other types of documents. It requires intelligent analysis of unstructured data in combination with the performance necessary to index and analyze millions of documents. The approach used by n3 and InterSystems makes possible big data solutions that can quickly and affordably be integrated into existing applications across industries and branches