About Siets.net project:

Team, History, News, Events

About Siets.net project and team

My name is Gints Ernestsons. I am a creator and owner of Siets.net, unique and scalable search engine and crawler software technology described on this website. I am also a co-founder and CTO of Contextprime SIA, a privately held European IT company in Riga, Latvia, developing and supporting 'SIETS' software products.

Origins of the brand name 'Siets'

Name 'Siets' in Latvian means 'Sieve' in English, a tool that people invented to sift grains of crop, to discover dirt-covered gold nuggets or to sort out other valuable material from a mix of content.

Siets software helps to discover valuable information among tons of data, working as a digital sieve for information.

Early years of Siets.net software development

The software was actively developed and supported by the team of 5-7 software engineers under my management, while I had a stake and CTO role in a privately held IT company Lursoft, co-founded by me in 1992.

Lursoft was regularly included in Fast50 and Fast500 nominations of the most rapidly growing Technology companies in Central Europe by Deloitte&Touche.

Our Siets team has developed a range of different Internet search and enterprise search solutions for Lursoft customers using Siets.net software in 2001-2005.

Powering search-driven products in spin-off companies

I have sold my stake in Lursoft in 2005 and retained ownership for Siets.net source code and software licensing, along the technology Web domain brand names Siets.net and Siets.lv.

After we mutually split Lursoft business in 2005, Lursoft other co-founders did stay in information services business in Latvia, while I continued to develop and to improve Siets.net technology in my own privately held or VC-backed businesses.

Since 2005 I co-founded several other business ventures licensing Siets.net search software technology for use in their products and services:

Siets a key search engine component in DBMS software from 2006

Document-oriented NoSQL database vendor with high-speed distributed ACID transactions (49,000 tps on a 30-node cluster) and unique Javascript-enabled JS/SQL API. Run as Cloud DBaaS service.

Since 2018 Clusterpoint DBMS product is being maintained and supported by Contextprime.

Siets licence from 2008

The second largest online subscription information service with tens of thousands of daily users, providing official data on all companies in Latvia, registration certificates, business activities, owners, officials, annual accounts, pledges, insolvencies, land and vehicle properties, paid taxes, anti-money laundering screening etc.

A privately held joint venture with German media house Heise Gruppe.

Siets licence from 2013

Imprimatur Capital VC backed advanced data visualization software vendor with unique Javascript-only cross-platform SDK software for building awesome interactive charts. Siets engine technology was used as a super-fast back-end data feed for Zoomcharts to showcase touch-screen oriented, fluid, clickable, very easy and intuitive to use and instantly responding visual charts demonstrations on Big data sets.

Please kindly see Zoomcharts DEMO gallery.

Today Zoomcharts is a well-known brand name in data visualization software market and the company customers are the world's largest Fortune 500 companies.

Siets licence from 2018

The latest co-founded by me IT company with strong focus on search solutions and services. Contextprime provides platform agnostic professional IT services to large local market customers (banks, telecoms, government), including software development and support for different commercial or open source search platforms. It has expert skills how to build search solutions using Siets Server, Siets Crawler, Clusterpoint DBMS, Lucene and ElasticSearch.

Business development strategy is to fully refocus on search software driven product business and start rapidly scaling the company business in the global market.

All of the above 4 ventures were initially set up as new startup companies with well-written business plans, with the entire tech and business team hired, and with founders leading the product development and the first market validation of products.

Two companies were VC backed.

Siets.net search engine software market readiness

For the last 15 years I was working along my engineering team members to improve and solidify Siets.net core search engine code.

Search engine software source code has about 170 000 written lines of code, with core software written in C/C++, tested and production proved in 365/24 mission-critical customer accounts.

Scalability tests for the software has been successfully performed using customer supplied Big data sets with up to 14 billion of document-type objects loaded for real-time search, analytics and reporting in a cluster of 100-servers.

Over the entire project life-time more than 120 man-years of engineering efforts were invested in the code development, documentation, prototyping and software quality testing in total.

I am using Siets search engine software in all of my other co-owned businesses since 2001. Among them, Firmas.lv depends on Siets software reliability and performance, running the entire business-critical online business information service built on this software technology instead of an SQL database platform.

Please kindly check section "Investors" of Siets.net website for new product investment opportunities, that could be powered by this unique search software technology.

Gints Ernestsons, Siets.net creator and chief architect


Siets.net History

Siets platform versions

Siets.net search engine software code has undergone through several version improvements since 2001.

Historically 4 different platform versions were released, to accommodate search engine software to the latest advances in computing CPU hardware architectures, as well as to support open ISO data standards in data management and software use in cloud computing:

The following platform versions are available for the software today:

  • 32-bit architecture XML data store version (since 2001)
  • 64-bit architecture XML data store version (since 2009)
  • Cloud JSON data store version (since 2015)
  • Docker-container JSON data store version (since 2020)

Siets products by other companies

Historically search engine technology based on Siets.net core software components has found its use in a range of other privately held or VC-backed company products, that I co-founded and did help to bring those products on the market:

Siets first introduction to market in 2002

Siets Server first version was introduced to the market as a software powering Latvian UNESCO Memory of the World heritage data corpus "Folksong Cabinet" (www.dainuskapis.lv) in 2002, as a novel search engine solution provided to "Institute of Literature, Folklore and Art" of the University of Latvia, by Lursoft.

Provisioning of instantly relevant search engine for millions of folk songs was a brilliant marketing method for Siets search engine technology.

Many people were eager to explore and search one of the probably best known by everyone information collections in the Latvian culture.

It got tons of positive publicity in media and TV and did serve as the technology DEMO for hundreds of thousands of Internet users.

Siets commercial success as an organic growth

Many other commercial customers soon also started to use Siets.net software as a search engine for their digital data properties, with Siets.net engineering team assisting and helping them to implement solutions.

Customer references can be found in the next section below: "News, Product Releases, Media Coverage".

Siets enabled product innovations by other companies

Unique Siets.net search engine technology powering massive amounts of data at blazing-fast speed, later enabled me to come up with new innovations in other software market segments and create new product companies.

In particular, Clusterpoint, a VC-backed Big data venture I did help to launch, using underlying search engine technology, was the first company in the world that implemented two key innovations for NoSQL database management software industry:

  • embedded Javascript based SQL query language
  • distributed high-performance ACID transactions

Clusterpoint Cloud database was used by more than 7000 signed up customer accounts from 2015-2017, with developer friendly hackatons organized all around the world in the biggest cities, and with more than 700 developer projects hosted on this database at the time.

Clusterpoint DBMS software product has its own licensing terms, as defined by Imprimatur Capital (www.impcap.com), UK, London, a trustee of intellectual property rights for this DBMS software (the company is being restructured today).

News, Product Releases, Media Coverage

Jan 05, 2020

A new ScalingSearch search engine Docker-based container product alpha-version launched.

ScalingSearch product has been built by Contextprime as a scaled down latest version search engine software component stripped off from all cloud data store software dependencies implemented into Clusterpoint DBMS latest version (version 4). Software was packaged for easy containerized deployment by any software developer in common virtual machine OS environment, using the software as a search tool in developer's software stack (similar to ElasticSearch) to build other search applications.


Feb 12, 2019

Clusterpoint Cloud Database demonstration prototype on a 100-server cluster is being run on behalf of Imprimatur Capital by Contextprime.

Contextprime provides maintenance of Telia data center hosted server hardware, administration and consulting about Cloud DBaaS (Database as a Service) software platform and demonstrates its Big data scalability for different business opportunities.


Jan 18, 2018

Siets.net search engine and crawler technology was licensed to Contextprime SIA business.

Contextprime SIA, co-founded by Siets.net founders and key engineering team members from Clusterpoint in January 2018, takes over technical support for all Siets search engine and Clusterpoint DBMS product customers in a newly created professional IT services company.


Nov 29, 2017

Clusterpoint software products entrusted to London based Imprimatur Capital VC fund (United Kingdom).

Clusterpoint Group companies in United Kingdom, USA and Latvia are being restructured by investors.


Dec 19, 2015

Clusterpoint DBMS Distributed ACID Transactions Whitepaper published.

Describes U.S. patent pending novel approach how to implement high-performance distributed ACID transactions in NoSQL data store.

Method was implemented in Clusterpoint DBMS latest version V4 software.


Aug 11, 2015

Clusterpoint enables Latvijas Talrunis to compete with GOOGLE in local search market.

Clusterpoint provided fully hosted NoSQL and search engine solution to the national market leading business directory and telephone book service. Complete migration from Oracle Server was accomplished, replacing SQL databases with NoSQL solution. Customer got significant savings in costs and in less complexity to operate a well-known web Internet business directory portals of the company: www.zl.lv, www.pilseta24.lv, www.abc.lv, www.medicine.lv and others.


Aug 08, 2015

Clusterpoint Cloud Database With Javascript/SQL Version 4 Released.

New API support released designed around SQL-like syntax with embedded Javascript code.

Cluster-wide computing engine implemented for super-fast server-side analytics using just SQL and Javascript.


Sep 19, 2014

Responsive, relevant and deeply personal online banking experience for everyone.

Solution delivered in cooperation with ZOOMCHARTS (www.zoomcharts.com). Instant visual simplicity for your online banking customers. Search, find and analyze your money flows at the speed of thought.


Oct 28, 2014

Clusterpark "GREP OF LOGS" (GOL) Product launched: Log Analytics And Search Platform.

Clusterpark"GREP OF LOGS" (GOL) product provides superchaged 'grep' utility performance rivaling Splunk performance for scalable machine generated log and events data records management, search and analytics at scale, while doing it at a fraction of Splunk cost.


Sep 13, 2013

Clusterpoint Monetizes Digital Assets At LETA, The National News Agency.

Driven by search, Latvian News Agency LETA business is providing paid subscription service targeted at media monitoring agencies, PR firms, researchers and analysts. It enables LETA to generate incremental revenues from historical media assets that would otherwise have been seen as pure archive material. Clusterpoint was used at the heart of this service since 2010 and was servicing thousands of subscribers issuing tens of thousands of search and retrieval requests against tens of millions of media assets every day. Online search service delivers consistent sub-second response times and runs on a cluster of 10 off-the-shelf hardware servers.


Mar 03, 2013

Global Internet Search Platform With Crawler Solution Delivered To A Customer.

Confidential contract prevents disclosure of customer. Software used for alerting customer employees about specific content changes on the Latvian national Internet for immediately actionable law-enforcement procedures.


Feb 02, 2012

Lattelecom BPO Call Center 1188 application solution launched.

Aggregates more than 35 data sources into a single searchable information lookup resource for Lattelecom 1188, the largest and oldest national call center service in Latvia. Complete workflow solution supplied, integrated with automatic telephone switching equipment, delivered as a workstation software for hundreds of call center 1188 operators, who are providing search-driven mission-critical 365/7/24 telephone and Internet information services across the entire country to tens of thousands of callers daily.


Nov 11, 2011

Clusterpoint Server released as a scalable DBMS software only product.

Downloadable software version of Clusterpoint Search Appliance without bundled hardware box, released as a pre-installed software package.


May 15, 2009

Network Traffic Surveillance System (NTSS) product launched built in joint cooperation with local network security company Ugunssiena IT (www.firewall.lv).

Captures all network traffic passing network perimeter, re-engineers all TCP/IP packet content back to full connections and OSI 7 objects, and indexes all reconstructed application objects (web pages, emails, docs, audio, video streams, DNS calls etc) for instant search. NTSS product is using high speed C-code developed network tapping software from Ugunssiena passively listening on Ethernet ports and using Clusterpoint DBMS built-in Siets.net search functionality as a search engine OEM component.


Feb 01, 2009

EU-funded EURECA research project E!Innobase approved for the development of 64-bit platform product.

64-bit architecture software platform development was initiated with funding from the experimental EU funded research project E!Innobase, applied in joint cooperation among Clusterpoint SIA, Institute of Mathematics and Informatics, University of Latvia, and Imprimatur Capital (UK, London).


Jul 01, 2008

Siets.net search engine technology was licensed to Firmas.lv business

Today Firmas.lv is a runner-up business information provider and the established brand name in Latvia, second to the market leader Lursoft in 2020. The company is operating in the same Deep Web information trading market segment as Lursoft.


Aug 21, 2006

Siets.net search engine technology was licensed to Clusterpoint SIA, a new VC-backed startup business in enterprise search market, for search appliance product packaging (with bundled Siets Server, Siets Enterprise Manager and Siets Crawler).


Dec 08, 2005

Siets.net search engine technology was licensed to Avedat SIA business

Privately held company fully owned by Gints Ernestsons for his intellectual property management.


May 12, 2005

Real-time alert functions added to Siets API with context triggers for developing content monitoring applications and software agents.

Riga, May 12, 2005.  Information technology company Lursoft IT (Lursoft) added automatic alerting functionality to it's flag ship software product - SIETS Server.

There are several new functions added to Siets API supporting creation of context sensitive triggers, which can be activated upon incoming or recent data updates by different user monitoring applications or scheduled software agents.

For example, application developers can now create triggers for filtering Siets database content  by keywords, phrases or Boolean expressions.  Each trigger created has its own ID code in Siets database.  Monitoring and  software agent applications can periodically examine any Siets documents against all established filters for specific Siets storage.  All matches found are returned as filter IDs to the user monitoring application.  Additionally in case of filter matching events Siets Server engine can execute the predefined script for event logging or messaging on the server side. 

In this way whole new set of applications can be developed such as subscription services for software agents, which periodically check recent document updates and send notification messages if the document updates contain words or phrases which match filters.

Developers can add, modify, delete triggers and examine any of the document against established triggers in the new API set.  This gives them great flexibility to activate monitoring functionality as frequently as necessary, or check for context triggers only upon new or recent updates.

The advantage of processing context filters on the Siets Server engine is a much better performance.  Siets context filters are processed in real-time, using Siets Server engine's generic full text index data, yielding about 10 to 100 times performance increase compared to the context filtering if done on an application server side accessing some database with separate SQL transactions for checking every content filter.  This gives opportunity to examine any document against tens of thousands of filters in sub second time.  This performance improvement becomes very important in large scale enterprise applications or Internet applications with tens of thousands of users having different individual needs for data monitoring.

New SIETS Server alerting functionality is described in Siets API Alerting Functions.


May 03, 2005

Web services .NET client sample code added for Siets Server Windows  users

Riga, May 03, 2005.  Information technology company Lursoft IT (Lursoft) added client sample code in Visual C++ .NET for its platform software product Siets Server.

Application developers programming for Web services which is popular standard on Windows platform, can now use standard Web services interface sample code to access Siets API functionality.  It contains all necessary SOAP and XML wrapping to access Siets Server to store documents or other data suing standard Web services interface.

Please take into account, that generic Siets API is more simple to use than .NET based Web services.  Otherwise Siets messaging principles follows the same concept of XML messaging over HTTP as in SOAP based .NET protocol.  Compared to .NET protocol, Siets API protocol is a bit less demanding - developers do not need to  spend time for dealing with document definition schemes.  In Siets environment API protocol is just a simple XML 1.0 messaging, which is more efficient in performance terms.

Yet for compatibility needs with an established industry standards Lursoft supplies also standard Web services based .NET interface sample code.  It can be used by developers for designing their own uses of any of Siets API command.

Support for .NET services also gives better integration option for those developers who work in heterogeneous Linux-Windows environments.  Windows based application servers can now fully exploit Linux based Siets search engine functionality using standard .NET tools provided with Microsoft Development kits.


Apr 15, 2005

XML-drill down features added for Siets API for building easy-to-navigate categorized result sets

Riga, Apr 15, 2005.  Information technology company Lursoft IT (Lursoft) added XML-drill down functionality to its flag ship software product Siets Server.

Application developers programming Web based user interfaces can now use a feature supported by Siets API called XML-drill down.  This feature returns categories and counts totals of results per each of the category for predefined meta tags <MENU>.  Totals per each menu-values driven category specified and matched with hits are returned together with every result set to the user application performing search queries.  This data can be used by user application to build easy to use  multi-level navigation tree of links and browsing systems for any search results.

This feature can substantially improve customer satisfaction with Web site search functionality and offer plenty of navigation choices.  Web entries for catalogs and sub catalogs can be build dynamically depending on what the user searches for and data present only in the matching result set.

For example, if the end user issues a query and finds 100 hits containing a full text term 'car', and car brand menu categories have been indexed as <MENU> items, then Siets Server API will return also all matching MENU value items and counters how many documents are found per each category in a search result, e.g., Ford 15, Mercedes 12, Toyota 10 etc.  This information can be used by user application to build Web navigation links to the following user search queries, which will narrow end-user choices of car information to respective brands only.   This navigation-by-results driven categories can be similarly applied to other navigation values, e.g., for car data the similar navigation can be built by color or types of cars etc.

In this way Siets Server engine helps to build a query specific top-to-bottom XML-based navigation trees called XML-drill down which contain only those categories and subcategories where some matching results are actually present.  For end-users it is extremely powerful option which does not require entering more specific search keywords - users can just click on the itemized MENU values to launch the next relevant search. It also helps to avoid using complete catalogs with hundred of categories to navigate, if the end user is interested just into a small subset from the all catalog categories.

In most cases users also do not need to build predefined catalogs or MENU values - any present XML tag can be used as MENU item. Siets Server will automatically collect all unique MENU tag values from existing XML documents and will count all relevant matches per each result set.

Description of new SIETS Server API functionality can be found here Siets Server XML-drill down features.


Apr 04, 2005

Automatic mirroring added for index redundancy and workload sharing among multiple servers

Information technology company Lursoft IT (Lursoft) added automatic mirroring functionality to it's flag ship software product - SIETS server.

Now all SIETS database updates sent to any server defined  as mirroring server in a configuration file will be automatically copied to all other mirroring servers present in the configuration list.

Most significant advantage is that mirroring of index updates should not be anymore done by search applications.  It has now been built into the SIETS engine and will be supported in all future versions of SIETS software.  This new feature will significantly reduce complexity of search application development for corporate environments.  Instead of programming application-level mirroring developers can now concentrate their efforts and time on other tasks.

New functionality will address primarily those enterprise customers who need to maintain multiple copies of data for mission critical business services where full text index should be available all the time.  Often hardware failures lead to unavailability of full text index due to its large size and problem to quickly restore tens or hundreds of gigabytes from a backup system.   Running two or more identical search severs in parallel can reduce this problem and shorten service unavailability time.  SIETS automatic mirroring among multiple designated servers can provide necessary redundancy without additional programming or administration overhead.

Built-in mirroring functionality on the SIETS engine will also enable more easy setup of SIETS server configurations where multiple  servers are used to share the total workload for all incoming search queries.  This feature will most benefit owners of large databases where search queries usually are slower due to gigantic index size and operations are close to the hardware memory limitations.  By splitting up all traffic of incoming search queries among multiple identical servers, total through output, each search query response time and overall system performance can be maintained at the necessary quality level.

Multiple server configurations in identical mirror configuration can also help to handle search traffic during a peak workloads.


Jan 13, 2005

Lursoft has summarized Siets.lv search portal statistics about most searched Internet content in 2004

Information technology company Lursoft IT (Lursoft) has summarized statistics about Latvian national search portal Siets.lv search requests in 2004. There were more than 13 million queries for different content performed by portal Siets.lv users in 2004. This number exceeds more than double the total number of search queries estimated by Lursoft one year ago.

Statistics data is published in the Internet address www.siets.lv/topi.

Statistics for 2004 was made using most popular words and phrases as search queries, grouped by categories. Every category has its own most frequently search terms, making TOPs: Companies, Periodicals, Celebrities, Events, Technologies, etc. Every TOP is being created by taking into account how many times search requests were made for respective objects in a category. Siets.lv users also can follow in real time what other people search.

Search portal Siets.lv become more and more popular among Latvian Internet users. Besides Internet content search, it allows to search in many other resources - Company Register, Newspapers (more than 2 million full text articles), Persons (more than 300 000 active persons in public and economy), Laws (more than 14 000 official legislation acts and regulations of Latvia), joint catalog of several large libraries and other databases. Today Siets.lv Internet search engine processes about 45-50 thousands search queries per day (see www.siets.lv/statistika").

Siets.lv portal was built using innovative search technology of the same name developed by Lursoft engineers - Siets Server. Pease see technology description and documentation on www.siets.net. Siets search technology is being used by Apollo portal, Bank of Latvia, Latvian National Library, Internet Web sites of several leading daily newspapers (Diena, Latvijas Avize, Neatkariga Rita Avize etc.), Latvian University and other organizations. Lursoft also started to export this software technology to customers worldwide: in USA, Sweden and other countries.


Jan 03, 2005

SIETS Enables Fast GPS Location Search and Distance Sorting

Information technology company Lursoft IT released new Siets Server software upgrade with built-in Numeric Range Search functionality.

Unlike other full text search products with rather limited numeric search functions Siets Server supports range searches for integers, floats and dates. Multiple numeric intervals for full text searches are also supported.

Unique feature in the latest Siets Server software version is Location Search With Distance Sorting for queries where numeric range intervals are specified. Result set can be sorted either by circle or square (or optional - so called taxi-distance) sorting algorithm.

This unique functionality allows to build different location search applications common in many business sectors. For example, Siets Server can be feed using GPS-navigation based geographic coordinates - longitude and latitude - and a full text query for all objects within specific distance radius in miles, kilometers or degrees.  Search engine will return all relevant results sorted according to the shortest distance to the center.

This functionality can be used by car GPS-navigation support systems, location based business directories, Internet search portals providing location based search using geographic coordinates, supply and logistics businesses, tourism and travel sector businesses, and many other businesses where information search is needed in respect of object's position.

Siets engine performance for location search is outstanding.  Engine was highly optimized for this specific type of searches by Lursoft. Siets engine can execute location based full text searches and distance sorting of results under 0.2 seconds in multi-million databases.  Overall through-output can be more than 1000 location search transactions per minute per one server.

Siets Server software latest version v3.3.3 can be downloaded from the Siets Server product site www.siets.net.


Oct 15, 2004

A Latvian Company Aims at Google's Territory - .PDF file

(from Baltic IT&T Review, published No. 3 (34), 2004)

Sep 12, 2004

Erstes Public Release des Siets Server

(from www.pro-linux.de, 12. Sep 2004)


Aug 31, 2004

First public release of SIETS Server software

Riga, August 31, 2004.  Information technology company Lursoft IT released the first public version of Siets Server software. The software is powerful high speed Linux based search engine developed by the company. It has been used for more than two years as Lursoft's internal search engine for various applications. The evaluation version of the Siets Server software can be downloaded from the Siets Server product site www.siets.net.

The software combines three emerging mainstream software technologies into one product: XML document storage, full text search and clustering. The engine runs as a database server and communicates with any client using simple XML messaging and http protocol. This open standard allows to integrate advanced search functionality into legacy applications and to use favorite programming languages to access Siets Server from other systems.

Siets Server search technology is used in the Latvian Internet search portal (www.siets.lv), the largest library of Latvian full text newspaper articles (www.news.lv), search services for the Bank of Latvia (www.bank.lv), search services for the national telecom operator Lattelekom portal (www.apollo.lv), national cultural heritage database Folk Song Cabinet of the Latvian University (www.dainuskapis.lv) and other information systems.

More information about Siets Server can be found on the Siets Server Web site www.siets.net.

About Lursoft. Since 2001 Lursoft was regularly included in Fast50 and Fast500 nominations of the most rapidly growing technology companies in Central Europe. This list is compiled every year by Deloitte&Touche. In 2003 Lursoft was also included among "All stars" ranking by Deloitte&Touche.  Lursoft is also a member of the European Business Register (EBR) and Latvian Information Technology and Telecommunication Association (LITTA). More information about Lursoft can be found on the company Web site www.lursoft.lv.


Jun 07, 2004


Jan 01, 2000

Siets Server Project Started

When Internet was becoming more and more popular at the end of the 90ies, along with many other Internet based services requiring public search functionality, Lursoft engineers lead by Gints Ernestsons returned to their own original concept of the FTS software based on the unique inverted index. Lursoft Fast Document Searching System prototype developed in 1993 using 4GL language provided basic design blueprints for the construction of the modern full-text search system.

Internet content search services launched by Lursoft required both scalability and new FTS functionality not present in legacy commercial FTS products such as language stemming and text analytics.

The resulting design concept of the Siets Server was created by Gints Ernestsons in 2000. It was based on three emerging technologies: FTS, XML and distributed computing through clustering of PC server computing resources.

For the maximum performance on PC hardware software the coding job for Siets Server search engine platform was started using C programming language.

Programming in C has delivered maximum indexing and search speeds and was portable across different operating system platforms.

Although core Siets search engine software was developed as a platform independent C code, for better cost controls open source software was also used as a supporting OS and web server platform tools - Linux and Apache.

Engineering team created the architecture of a completely new system for Siets Server core engine which operated as an XML native database. In such a system all objects were stored in the server's database storage formatted as simple XML documents.

Client-server communications between Siets platform and client applications were using simple XML based messaging protocol and user friendly safe http/https based networking, supported by majority of other Internet software platforms (including web browsers).

Open XML based architecture and HTTP messaging protocol allowed to use Siets server FTS functionality from different client environments or scripting languages.

In less than 3 years time Siets Server basic engineering framework was created for Siets Server: open client-server XML document messaging database with built-in FTS functionality.

Clustering support was also added to Siets Server by design to take advantage of abundant low-cost commodity hardware.

Clustering support allowed to distribute data and workload among many networked Siets Server computers to achieve linear performance scalability for indexing and at the same time maintain the same very high search speeds across the whole distributed database.

There were numerous mathematical methods and clever data processing algorithms built in the Siets code during the initial development and testing period of Siets Server. Hardware advances were taken into account when optimizing software code to match computer memory size and limit unnecessary repeated disc storage interactions.

Enterprise level support for the multilingual, multiprocessor and multi-computer operation modes by running several Siets Server instances in parallel was implemented as standard features of the Siets Server and its HTTP based API.


From 1995

Using Commercial SQL-bound FTS Solutions

In 1995-1999 Lursoft invested into the infrastructure of creating and running SQL relation databases.

To process FTS search requests for the Lursoft Newspaper Library (a collection of millions of full text articles of Latvian central and regional press), a commercial full-text search tool (Verity search Engine) was purchased by Lursoft in 1996, which was based on Sybase SQL database server and operated as the additional full-text indexing and searching module to this database.

Although this solution generally worked, it was heavily integrated with the core SQL system and thus very difficult to maintain and administer. It also lacked real-time index updates, causing data and search results inconsistencies in Lursoft online web subscription services started in 1995.

SQL-bound FTS platform quickly became very slow for Lursoft's needs in its growing Internet online search and subscription business.

At the end of 1999 it was clear to the company management that Verity search engine did not meet scalability requirements.


From 1993

Fast Document Search System Project

In 1993 Lursoft started to develop a prototype of the text data search system on the basis of the inverted index, using the FoxPro database management system as the development environment.

In this environment documents were divided into separate words and saved in the system of linked matrix, creating inverted index of all words and their positions in text. This system had all the elements required for a fast full-text search platforms, including inverse index building modules, algorithms of FTS search and support of stemming in national languages with different endings or prefixes.

The prototype of the system was developed in 1993. A graduation diploma work of a Lursoft engineer Ugis Koops under Gints Ernestsons couching and mentoring as CTO was also made.

At the faculty of Automatics and Computer Technology at Riga's Technical University the project "Fast Document Searching System" was successfully presented as the excellent engineering graduation project.

Due to FoxPro's architectural limitations, being a database scripting language and interpreting 4GL code operators overall performance of this prototype was not very high. This pilot software was not implemented in commercial FTS software solutions. However the project was used for internal Lursoft desktop databases to search more than 40,000 legislation documents (laws, regulations) with sub-second response times. It was a good test bed to test future Siets.net FTS concepts, algorithms and prototyping a data model at high level programming language.


From 1992

Full Text Search Software Concept Origins

The idea to develop software for very fast full-text search engine (FTS) first appeared to Gints Ernestsons in 1992, when Lursoft was founded as a company. By that time IT sector using PC technologies was dominated mainly by DOS and Novell. Dbase, Clipper, Paradox and FoxPro relational databases were widely used.

During the development of the automated system of the National Company Register for the Government of Latvia, Lursoft engineers realized the need to have a tool for search of information in textual data. At the beginning, there were company articles of association and other formal application documents which the Company Register wished to process.

Everyone who has worked with DBF type relation data files knows that it was not exactly the best format needed for the requirements of the storage of documents of this type. In the FoxPro database system chosen by Lursoft all the information had to be strictly divided into tables and fields. The largest document that could be stored in DBF format at that time could not exceed 64 kilobytes. The architecture of the relational databases provides fast information retrieval only if using strictly defined indexes, but can not do fast searches in the textual data of rather free format. At the time, when the fastest processors were 25Mhz and 33Mhz, it was very difficult and time consuming to find required information among thousands of textual document files residing on PC hard disk. Frequently, simple context based searches took tens of minutes, but in large archives of information it could take even several hours.

It was necessary to develop a better solution for FTS.

Lursoft initially choose the same FoxPro database management system to develop the tool for FTS system in order to do FTS search within textual information.