Instant update is a significant improvement on incremental updates. While incremental updates can achieve data update cycles of just a few minutes, with instant update the main elements of modified data are immediately searchable due to the instant feed.

In addition to the extremely error-tolerant multi-field search and single-field search in structured data, the latest version of Exorbyte also offers a flexible form search (FlexForm) that combines the best of both features: users can enter queries in multiple fields, but MatchMaker will apply these to the data dynamically, based on the best “interpretation” of the query. Forename and surname elements may be entered in the wrong fields. An address field can also be automatically split into town, street and postcode. And all of this is possible without elaborate programming, but just by using simple configurations and in the short processing times you expect.

Using appropriate facets / filters customers can easily refine search results in the dynamic navigation in order to quickly find the desired product. If the facet values are only contained in unstructured texts, they must be extracted and normalized to allow use as a facet. MatchMaker now has a built-in facet extractor that learns in a semi-supervised manner to extract facets from sources such as web-pages, abstracts or product descriptions. Any facet, such as color, size, rating and other characteristics, can be recognized and extracted. The results can be automatically added to the search index to enrich the data.

Crawling the web or an intranet site is used to collect data together for indexing. MatchMaker now tightly integrates the well-known open-source crawler Nutch. Bringing unstructured data into MatchMaker is made simple through the many connectors that Nutch provides and we provide an easy-to-use configuration and maintenance module for Nutch directly in exTractor, the MatchMaker ETL tool. These advantages of Nutch are now integrated in MatchMaker:

• Simultaneous crawling of multiple URLs due to high performance
• Maintaining all, also varying security standards while searching open and secure sites due to individual fine tuning
• Extensive adjusting due to JavaScript support
• Indexing about 500 data formats due to integrated document conversion
• Adding open source or individual plug-ins due to available interfaces

Multi-Level Search is a powerful search tool that makes search queries significantly more flexible. With Multi-Level Search, it is possible to search within defined hierarchical levels. When searching for particular towns in a person’s database, for example, towns are filtered out on the hierarchical level based on all town or street names that match the search query, with no reference to individuals. But now queries such as “give me all streets where a Mr. Schmitz and a Mr. Maier live” are also possible.

MatchMaker now also supports data extraction from XML files. The configuration process is displayed graphically, making it easy to use. This feature is particularly helpful when used with multi-level search.

The configuration wizard statistically analyzes the structure of data to be integrated in MatchMaker. On the basis of this analysis, the wizard offers useful MatchMaker configuration suggestions, making it easier to configure a project.

This feature is particularly beneficial with very large data volumes and short update cycles. Alongside the normal incremental update, the “distributed build” feature allows indexing to be distributed to multiple processors and/or machines, as well as to be partly parallelized. Parallel processing considerably accelerates the indexing process.

This new feature automatically extracts color values from pictures such as product images and then allows searches by color name (“blue”), color range (“reddish”) and RGB or HSV values. This makes it possible to search for images with similar colors, such as for all “red” or “reddish” products.

This add-on allows exLog, MatchMaker’s log server, to be connected directly to external monitoring systems such as Nagios, enabling log files to be monitored centrally in a monitoring system.

The new exPlain tool analyzes the functionality of difficult-to-understand algorithms, such as the Levenshtein algorithm, and displays the interdependencies in a graphical format. Primarily designed as a test environment and learning tool, exPlain helps you to better understand MatchMaker’s structures and the effects of particular search configurations.

• Designed and developed for use in the cloud
• Practically unlimited and automatically scalable
• Highly available, secure, reliable and effective
• Unlimited multi-tenancy
• Individual point-and-click customization
• Easier connection to existing systems via API interfaces
• No outlay for hardware, software or updates
• No need to worry about server infrastructure administration and maintenance
• Shorter implementation time for faster use in live systems

• Single and multi-byte character encoding
• Umlauts, upper and lower case, special characters

• Extraction library
• Linguistic method and clustering
• Phrase extraction
• Transliteration
• Morphology

• Damerau-Levenshtein and related algorithms
• Longest common subsequence
• Wildcards, glob-style queries and regular terms
• Finite automata (DFA)
• Hit qualities

• Free transposition of words
• Consideration of word relevance
• Multi-word scanner
• Displacement of word elements, block edit calculation

• Phrase recognition (even with errors in individual words)
• Exclusion terms
• Word completion and optional words
• Approximative or exact word-based queries
• Word combination and splitting
• Single word aliases and bias weighting
• Individual and automated skip and stop words
• Wildcard, prefix, infix and suffix search

• Soundex, Metaphone, Kölner Phonetik
• Exorbyte phonetics
• Freely programmable phonetics
• Formal similarities

• Strategies – combination of algorithms in fields
• Cross-field index for fast query analysis
• Alternative indexing

• Approximative date matching
• Ratios and number ranges
• Flag and option attributes
• Geometric distances
• Viewing and authorization concept

• Special indexing methods
• Matching based on the Levenshtein algorithm
• Wildcard integration
• Allows additional similarity tables to be integrated

More about MatchMaker OCR

• Weighting and confidence for individual fields
• Approximative intersection
• Post-evaluation logic for low similarity
• Boolean logic
• Free multi-field search

• Fast counting, navigator creation
• Sorting (language-dependent collation)
• Merging, splitting, deduplication

• Increased error tolerance for individual terms
• Entry relevance boosting
• General field preference

• Local aliases
• Global aliases
• Systematic aliases

• Data import from ODBC, CSV and other sources
• MMI programming interface in multiple languages
• Server-side scripting in Java and TCL
• Console management for remote control and OEM integration
• Java administration interface (JMMI)
• Analyzable reporting format

• Analysis tools
• Configuration tool
• Monitoring tool
• Statistical data
• Centralized logging
• Error reporting
• Reliability safeguards
• Centralized management of distributed systems

• Data parallelization
• Automatic adjustment to server load
• Query parallelization
• Data update
• Incremental data update
• Alternative search profiles
• Windows, Linux and 32 & 64-bit Solaris support

Many companies face the daily challenge of managing incoming mail that sometimes amounts to thousands of documents. Incoming forms and documents are digitized, analyzed with OCR (text recognition) and allocated to the relevant departments and applications. However, even the best scanners and recognition technology are unable to conclusively recognize documents and assign database entries 100%. Recognition and assignment errors create a considerable need for cost-intensive manual processing, and this is exactly where Exorbyte’s MatchMaker comes in. In addition to reading and spelling errors, typing mistakes or incorrectly separated words, MatchMaker even takes into account terms with related content and the context – e.g. file number or customer ID. As a result, the majority of incorrect documents can be assigned to the correct database entries, and manual post-processing can be significantly reduced. This produces a lasting improvement of business processes and reduction in costs.

Practical example: Allianz AG

• Manual post-processing significantly reduced
• Data quality increased
• Easily integrated into existing infrastructure
• Flexible settings for matching criteria
• Top performance

The search function is the central access point to CRM systems. Even the most powerful and most flexible CRM system can only perform at its full potential if all information can be found quickly by its users.
An intelligent search should therefore be error-tolerant and find relevant results even if a search term is entered incorrectly, flag duplicates and work effectively even with very large data volumes. MatchMaker does exactly that. Exorbyte’s intelligent CRM search provides users with the right names, numbers or other information in milliseconds. It is even possible to combine multiple search terms, such as “workshop premium customer Bielefeld”. MatchMaker also includes an intelligent auto-suggest, which provides relevant hits as the term is entered – even if the user mistypes or enters an inaccurate search term! Even in auto-suggest, only data appropriate to the user’s role will be displayed at any time. MatchMaker can be easily integrated into any environment, and can be adapted to suit your requirements. And all of this is also possible in the cloud, such as with the www.genialotel.de telephone directory.
Exorbyte’s CRM search is particularly worthwhile for CRM system suppliers, resellers and integrators: a high-performance, error-tolerant search represents significant added value for customers – this can be a crucial competitive advantage.

Practical example: Exorbyte’s CRM search in Projektron BCS

• Fast search across all BCS areas
• Error tolerance even in auto-suggest
• Easily integrated in just a few hours
• Find everything lightning-fast

Around 70,000 trademarks were registered in Germany in 2011. Before registration comes research to ensure that the prospective trademark will not affect or infringe the rights of an existing trademark. An accurate search is expedient for this, particularly if Nice classes can be entered in the search. In addition to exact hits, Exorbyte’s search and matching engine also finds and displays similarities in the immediate vicinity of the trademark. In this way, Exorbyte’s trademark search ensures that a new registration does not affect existing trademark rights. Conversely, it allows owners of existing registered trademarks to reliably protect their rights. Data on around 780,000 trademarks registered in Germany (as of 2011) is analyzed and searched in a flash, with results available within milliseconds even when searching all international data (approx. 80 million datasets). At the same time, it is important that users can depend on the search results being complete.

Practical example: www.polymark.de (bck GmbH)

• 3 million datasets
• Approx. 2,500 search queries per day on the portal
• Incremental update
• Efficient resource management
• Top performance

Websites are a significant sales channel for many companies, with customers accessing information and placing orders every day. Sometimes customers may register multiple times when ordering or signing a contract, and this leads to inconvenient duplicates. A more serious issue involves fraudsters who register multiple times, for example, to take advantage of discount offers. Using MatchMaker allows you to effectively prevent these duplicates and thus cases of fraud. To do this, MatchMaker checks whether the address entered actually exists by comparing the customer data with a street database during the registration. It also checks whether the customer is already recorded in the customer database. This deduplication requires an error tolerance that can be adjusted as needed, as offered by MatchMaker. This is the only way to ensure that each customer is only recorded once. MatchMaker also updates the database index within seconds so that duplicate registrations can be immediately detected. This systematically reduces cases of fraud and ensures the effectiveness of customer checks.

Practical example: congstar GmbH

• Fuzzy multi-field matching
• Data volume: 1.2 million records
• Response times: <10ms
• Incremental update
• Unlimited scalability
• High-performance error tolerance, adjustable as needed

Data quality has become an important and measurable economic factor in many areas of business and public service. Company master data is accessed by many different users and applications, including for direct marketing, automated incoming mail, case processing and automatic status queries. All departments suffer if master data is not clean. Exorbyte’s Master Data Server (MDS) demonstrably improves the quality of address data and thus the level of processing automation. What is unique about MDS is that it compares every query with other entries in the reference database to detect duplicates.

Practical example: BKK billing center, Emmendingen

• Number of queries: just under 4 million per month
• Fuzzy multi-field matching
• Improved master data search to check billing documents for over 150 health insurance schemes with more than seven million members
• Incremental update

Companies with Internet or Intranet websites as an important information or sales medium should ensure that users can find the information they require as quickly and easily as possible. Considerable importance should be attached to the search feature as a key website control. Google and the like have established certain expectations among users. They expect hits on a website search to be as relevant as possible, even with typos or inaccurate entries.
The full-text site search feature allows content throughout an entire website – including files in PDF, Word, PowerPoint and other formats – to be found using the site’s search box. Hits are displayed on the results page with the surrounding text passage and with the search terms visually highlighted. There are several options for capturing website content, including crawlers with individual settings to determine what should be crawled and when. This allows document types to be set, for instance, or non-changing standard text to be excluded.

Practical example: WMF

• Structured data and full text
• From Magento and Typo3
• Intelligent search with dynamic navigation

According to many studies, the search feature is the key control for online shops, making it particularly important to sales. Up to 80% of shoppers use the search function almost exclusively to find a product – and this figure is rising. Exorbyte Commerce Search is our intelligent, error-tolerant in-shop search with powerful, sales-boosting features. Relevant suggestions are offered to shop visitors while they are typing. Unexpected inputs or typos, singular and plural forms, synonyms or compounds are not an issue. Exorbyte Commerce Search finds all relevant entries and displays the correct terms. With the help of dynamic navigation, users can easily restrict results by category, brand, price or other shop-specific filters. The search also has a range of informative reporting and analysis tools that clearly show the shop operator what is searched for and how, or what is not found.
This makes it considerably easier for online shop visitors to find and buy products. User-friendliness is drastically increased because the search query does not need to perfectly match the database entries. The use of an intelligent product search increases the conversion rate by up to 80% and sales by up to 30%.

Practical example: www.elektro4000.de

• Approx. 350,000 items
• Approx. 20,000 search queries per month
• Search suggestions in milliseconds
• Error-tolerant auto-suggest
• Customizable filter navigation