Version 1.1 coming

Frank Bergmann — Fri, 11 Jun 2010 01:26:56 -0000

Release: ETA Summer 2010 Goal: provide a workable (turn-key) TM backend for SMB environment Targets - maintain signature of current API functions - extend API - user management - improve performance (matches < 1 second) - improve installation proceedure - scripted installation - provide default roles and users - improve documentation - provide basic postgres tuning guidelines - provide basic guide lines for managing users and roles Caveats - Bias - translation for software environments - testing will be done with Virtaal client (others are available )

Common Sense Advisory on TinyTM

Frank Bergmann — Fri, 13 Jun 2008 17:15:41 -0000

Hi,

http://globalwatchtower.com/2008/06/12/new-breed-translation-memory/

Common Sense Advisory, the leading market research company in the translation space writes on TinyTM:

> "[...] From a feature/function perspective, this
> CAT tool should meet the needs of many
> translators.
> From a GNU GPL licensing perspective, it will
> hit the low- or no-cost price point that is
> required for broadening the use of TM among
> translators. We’ll write more about TinyTM in
> an upcoming Quick Take on translation memory.

MetaTexis for TinyTM Released

Frank Bergmann — Thu, 05 Jun 2008 20:24:56 -0000

Barcelona, June 5th: MetaTexis (http://www.metatexis.com/) is a commercial (closed-source) translation memory vendor. MetaTexis has just release a variant of its MS-Word client with support for the TinyTM server. This way, users with the TinyTM open-source client and users with the MetaTexis closed-source client can connect to the same server and work together.

The TinyTM team believes that this release is an important step for TinyTM. Many organizations such as large companies or international organizations don't care too much about the price. Instead, they are looking for a product supporting an open standard with high usability and professional support.

Also, this step validates the basic TinyTM architecture and demonstrates that it is easy to integrate other systems with the TinyTM server, providing a suitable ground for other companies and projects to participate in this emerging ecosystem.

About MetaTexis:

MetaTexis (http://www.metatexis.com/) is a commercial (closed-source) translation memory vendor. Its MetaTexis translation memory is an integrated translation environment for various variants of MS-Word.
Contact: Hermann Bruns, bruns@metatexis.com.

About TinyTM:

TinyTM is a small but powerful open-source translation memory (TM). It features a client-server architecture, fuzzy matching and both single-user and multi-user (networked or over the Internet) operations.

At the moment, the client environment consists of a Macro for MS-Word. This MS-Word client is meant just to be the very first environment. More client environments will follow. The server environment consists of a PostgreSQL database including a built-in a full-text search engine in order to speed up the search for fuzzy matches.
Contact: Frank Bergmann, frank.bergmann@project-open.com

TinyTM V0.1 Released - Available for Download

Frank Bergmann — Thu, 05 Jun 2008 20:05:35 -0000

Hi,

This is the very first release of TinyTM. Version V0.1 is now available from download at http://sourceforge.net/projects/tinytm/ -> Download.

Scope of Version V0.1:

V0.1 is a "developer version" targeted towards users with certain database and system administration skills. For details please see the TinyTM homepage at http://tinytm.sourceforge.net/.

About TinyTM:

Announcing the TinyTM Translation Memory

Frank Bergmann — Tue, 26 Feb 2008 22:37:19 -0000

Hi,

TinyTM is a minimalistic yet powerful translation memory (TM) as open-source software. It features a number of features such as a client for MS-Word, fuzzy matching and both local and network operations.

The code base version of TinyTM currently consists of only 1000 lines of code. We believe that the limited complexity should encourage many developers to customize TinyTM to their needs and to develop their own open-source TM solutions.

The client side:

The current client environment consists of a Macro for MS-Word. When activated, the macro takes the current segment and compares it against the server in order to retrieve a number of fuzzy matches. The user can review the found fuzzy matches, select one as the base for his translation and commit his translation to the server. During the
commit process, the translated segment is enriched by meta-information commonly know as "tags". This way the translator can distinguish between different subject areas and effectively implement "cascading" TMs.

This MS-Word client environment is meant just to be just the first environment. We started with this "platform" because a) MS-Word is _the_ preferred translation environment for translators and b) because MS-Word is actually relatively easy to extend.

We expect a number of additional environments to emerge. In particular, we believe that existing applications such as KBabel or OmegaT could be modified to work with the server back end.

The server side:

The server environment consists of a PostgreSQL database plus a built-in a full-text search engine in order to speed up the search for fuzzy matches. This combination allows to maintain very large databases of translation while maintaining fast access and sophisticated ranking of fuzzy matches.

Access to the server environment is currently possible via the open ODBC protocol. We soon expect middleware to appear to wrap ODBC calls into XML-RPC, SOAP and other web service protocols.

The protocol:

The "protocol" refers to the definition of the data that are exchanged between the client and the server and their precise meaning. We have defined the protocol in such a way that it is easy to extend.

At the moment the protocol consists of three operations only:
- Retrieving the list of available tags
- Retrieving a list of ranked fuzzy matches for a given segment and
- Committing a segment to the server together with its meta-information

We expect the list of protocol operations to increase with time the maturing of TinyTM. However, these three calls probably represent the minimum that any client would have to support.

Fuzzy-Match performance:

One of the most important features of TinyTM is the scalable full-text engine that allows indexing millions of segments while still responding to queries within fractions of a second.

TinyTM does not implement this functionality by itself as many "leading" industry tools do. Instead, TinyTM relies on the built-in PostgreSQL full-text search engine "TSearch2" to do the heavy lifting. TinyTM only filters the returned results and ranks them according to adherence to the original segment and the specified Tags.

Recent posts to news

Version 1.1 coming

Common Sense Advisory on TinyTM

MetaTexis for TinyTM Released

TinyTM V0.1 Released - Available for Download

Announcing the TinyTM Translation Memory