The core functionality of TRS Database Server full text database system is an implementation of searching and unified storing the full text. The so-called full text retrieval is fast and accurate acquiring the information within the massive unstructured information.
The TRS Database Server pursues the balance of 100% recall ratio and high precision ratio, outputting the result can be arranged in order by field value and degree of correlation, supporting the taxonomy as well.
To address the art of state requirements of Internet, E-government, electronic commerce applications, etc, the TRS Database Server developed Native- XML processing capability to handle the semi-structured information,
|
| |
| |
TRS Database Server Overview |
| Consummate management of unstructured information |
Unified management of unstructured and structured data
The system supports various documentary formats with storing, indexing and retrieving functionality such as TEXT, HTML, RTF, MS OFFICE, PDF, S2/PS2/PS, MARC, etc. The multimedia stream data type and often used data, such as date, char, and numeric type are supported as well. |
| |
Holographic indexing for XML data
The TRS Database Server realizes XML as a complete and single data type, and store and processes XML file directly rather than storing it after decomposing. The tags and the attribute of the tags can be searched and accessed |
| |
Supporting multilingual and multi character sets
The system supports Chinese, English, French, German, Russian, Arabic language, etc. GB2312/ GBK/GB18030 codes, the BIG5 codes, UTF8 code are supported as well. |
| |
Highly cost efficient index compresses, implements very low rate of index space expansion
rate of index space expansion is at-0.2~~1.0 |
| |
| High efficiency , Intelligence and Real time |
Trade-off between recall and precision ratio
Multiple indexing policies optimize the recall and precision. According to the characteristic of Chinese, Indexing data based on characters and words, or both characters and words, can guarantee the certain level of recall and precision. The maintenance of theme dictionary can also improve the precision. |
| |
Intelligent searching experiences based on advanced natural language processing technology
Participle dictionary: system maintains the theme dictionary, thesaurus dictionary, antonym dictionary, stop word dictionary on demand
Participle Rules Library: ?established a lot of rules of eliminating the ambiguity to improve the precision of participle to lead high precision ratio of information retrieval.
Natural language search : segment the a sentence into the key words, output the query results in the order of the degree of correlation based on the position and the frequency of the key words. |
| |
Index dynamic updating to address the needs of transactional application
To meet the requirements of real time enterprise search, the system will synchronize to update the index when the records in RDBM are added, deleted and changed? |
| |
Full dimensional searching methods and multiple presentation style
Simple searching methods based on character, word, sentence; supporting 48 boolean operator provides the functions of position searching, second searching, progressive searching, etyma searching to address the needs of professional application such as patent, business intelligence and R&D.
The searching result could be put in order based on the relevancy , one or more featured filed value. The taxonomy, statistic and key-words marking are supported as well. |
| |
Multi-threads, high concurrency, fast response for large scale size of data volume
The system is designed on a multi-threads basis, supporting SMP architecture and database replication and mirroring and employing the perfect CACHE technology tp improve the parallel performance. Also, the system employs advanced indexing and querying algorithm and parallel query to achieve the goal of responding the searching request in milliseconds when the size of data at the level of gigabytes. |
| |
Supporting multiple language searching for internationalization
The client supports the following three character sets: GB2312/GBK/GB18030BIG5UTF8 for the multi-language application development. The system also supports more than 10 languages word segmentation and etyma processing such as English, Russian Arabic, etc. |
| |
| Powerful Enterprise application infrastructure architecture |
Distributed Cluster Architecture meets the increase of the size of data
Take advantage of TRS cluster server, TRS database server stores the data in the different physical servers to implement distributed data management and searching. The TRS database can be scaled out on demand when the data increases to meet the requirements of searching for large size of data. |
| |
Load balancing mechanism can guarantee the high performance under the condition of high concurrency
Under the condition of high concurrency, the data will be deployed multiple TRS Database Servers. The requests of searching can be dispatched dynamically to improve the load capacity of the server. The TRS Database server could be added on demand while the size of data increases. |
| |
| Security, reliability, maturity and openness |
Provides the four level access control mechanism, online backup and multiClevel? privilege management
Provides the four level access control mechanism: system level, database level, record level and field level
Database and all system information online backup
The privilege management with multi-level supporting user groups and Group DBA to gear to the complicated enterprise applications |
| |
Open and scalable architecture and standard compliance APIs
Supporting multi-platforms , mainstream OS, middleware and Web Server, Supporting 32bits/64bits OS Platform. The size of data in single database can be up to 2G or 4G.
Extensible Plug-in mechanism can support the data filtering. It gives users customized capabilities of implementing the filtering and encryption for the various documentation including Microsoft Office. The system will index the documents automatically.
The standard C, COM, ADO, JAVABEAN application interfaces are provided to meet various needs for application development. |
| |
Proven practicability and reliability
More than 3000 enterprise level customers have chosen TRS Database Server to build their own IT infrastructure. TRS Database Server has 70% share in the market of unstructured database. |
| |
| |
TRS Database Server Cluster Architecture
TRS Database Cluster Server is a distributed management system which deploying TRS Database Server on the multiple physical servers. The cluster architecture support distributing data on multiple servers and load balancing to address the requirements of high concurrency and large scale of data. |
Figure1? The architecture of TRS Database Cluster
|
Load balance within database group
The TRS database cluster dispatch the database server within database group unified. A request will be sent to one of DB server in DB group and one DB group consists at least one DB server. |
| |
Distribution searching will be implemented among the TRS DB Groups
The request from user will be sent to part of or all of DB groups based on the situation of distribution of the target objects the request contains. TRS Database cluster server will merge the data sets from all DB groups and feedback to users. |
| |
Taking advantage the architecture of cluster, the following goals will be achieved
Distributed searching and scale out on demand for the large scale data
High concurrent users requests can be handled with high performance
No single point failure and high reliable application development |
| |
Features of TRS DB Server
Managing heterogeneous resources unified, federated searching for unstructured and structured data
Usually, the enterprise users need to manage and search the unstructured and structured data unified to get more accurate information and mine the value of the information by information classification and association. For example, enterprise users need to federated search the office documentation and sales data, government needs to search the policies, laws and information crawled from the Internet. TRS Database Server can store and manage not only the web pages, Office documentation but also the relational structured data and support multiple data type and native XML so that it implements federated search for structured and unstructured data. |
| |
Trade-off between recall and precision, more advantage than traditional search engine
Most Internet search technology employs the pre-calculate algorithm and can not feedback the result in full and accurate. The commercial page rank mechanism also reduces the precision of results. Unlike Internet search engine, business oriented enterprise search has very high demand for recall ratio and precision ratio. TRS DB server provides various indexing policies to meet enterprise search demand. |
| |
Dynamic indexing and real time search, enterprise business information oriented search engine
As part of core infrastructure, enterprise users have very high demand on real time and efficiency search engine. TRS DB Server employs real time and dynamic indexing technology to update and synchronize index while the original data stored in RDMS are added, deleted, and changed rather than rebuild the whole index so that the user can search and get the latest data. |
| |
High performance and reliable enterprise class search engine
XinHua News Agency and Chinese Central Government Website deployed TRS Database cluster server due to the high performance, scalability, reliability of the product. The system can handle terabytes data and high concurrent users requests based on certain SLA. |
| |
| |
Operating Environment |
| TRS Database Server |
Unix(Solaris2.6/7,AIX4.3/5.1,HP-UX11.0)
Microsoft Windows NT/2000/2003
Linux2.4
Memory128M; Recommendations256M |
| |
|
| |
| TRS Database Cluster |
Microsoft Windows NT/2000/2003
Linux2.4
Memory128M; Recommendations256M |
| |
|
| |
| TRS Admin |
Microsoft Windows NT/2000/2003/XP
Memory128M; Recommendations256M |
| |
|
| |
| TRS CAPI |
Unix(Solaris2.6/7,AIX4.3/5.1,HP-UX11.0)
Microsoft Windows NT/2000/2003/XP
Linux2.4
Memory128M; Recommendations256M |
| |
|
| |
| TRS ADO |
Microsoft Windows NT/2000/2003/XP 128M; 256M |
| |
|
| |
| TRS JavaBeans |
Unix(Solaris2.6/7,AIX4.3/5.1,HP-UX11.0)
Microsoft Windows NT/2000/2003/XP
Linux2.4
Memory128M; Recommendations256M |
| |
|
| |
| TRS .Net API |
Microsoft Windows NT/2000/2003/XP
Memory128M; Recommendations256M |
| |
|
| |
| TRS PAPI |
Microsoft Windows NT/2000/2003/XP
Memory128M; Recommendations256M |
| |
|
