Tilde Custom Machine Translation

Learn more
Tilde MT license
This document is a Data Provider Agreement (hereinafter referred to as "Agreement") is concluded by and between the Tilde MT Consortium formed under the Tilde MT Consortium Agreement concluded on 01 March 2010, (hereinafter referred to as "Tilde MT Consortium") and You. Please read the terms, relating to the project entitled "Platform for Online Sharing of Training Data and Building User Tailored MT", in short – "Tilde MT" (hereinafter referred to as "Tilde MT"), carefully. They apply to the platform named above, any updates, supplements, and support services for this platform, unless other terms accompany those items. If so, those terms apply.
By uploading data, you accept these terms. If you do not accept them, do not upload.
1. Definitions

API means Application Programming Interface

Corpora means non-downloadable data uploaded to the Resource Repository

Data means the collections of texts that Data Provider provides to Tilde MT Services.

Tilde MT Database means data storage environment where data relevant to the Tilde MT system is stored including user account, and other sensitive information.

Tilde MT General Terms and Conditions means "General Terms and Conditions" document available on the http://www.letsmt.eu web site which governs the usage of the system.

Tilde MT Services means the following, but not limited to, data upload, storage and processing (including deletion), SMT system definition, training, running, and performing operations using supported application programming interfaces (API).

Tilde MT Service Subscriber means a legal entity using the Tilde MT Services authorized to define a number of users in the Tilde MT platform with different responsibilities, e.g. training data upload, training of systems and translations.

Tilde MT User means one of the following 5 available types of users:

  • Anonymous User (AU) includes any user from the Internet who accesses only the Public Data and public SMT Systems for evaluation, without authentication.
  • Registered User (RU) includes any user authorized to use trained SMT Systems. The RU is authorized to access only the Public Data and public SMT Systems and the SMT Systems of their respective Tilde MT Service Subscriber. The RU can use SMT systems both through the Tilde MT web interface and the API.
  • Registered Power User (PU) includes any user authorized to manage the Training Data, training tasks and trained SMT Systems of their respective Tilde MT Service Subscriber, including uploading data, browsing and reviewing details of corpora, editing metadata and deleting the corpora owned by their respective Tilde MT Service Subscriber.
  • Tilde MT Service Subscriber Administrator (TA) includes any user having full control over their respective Tilde MT Service Subscriber data set. The TA creates that data set, manages users of the Tilde MT Service Subscriber, etc. The TA is authorized to manage the Training Data, training tasks and trained SMT Systems of the respective Tilde MT Service Subscriber, including uploading corpora, browsing and reviewing details of corpora, editing metadata and deleting the corpora owned by their respective Tilde MT Service Subscriber, as well as grant these rights to any respective Registered Power User(s).
  • Tilde MT System Administrator (SA) is appointed by the Tilde MT Consortium and has maximum available rights to administer the Tilde MT Database. The SA is manages the Tilde MT Service Subscribers, all users, Training Data, training tasks, trained SMT Systems, including uploading corpora, browsing and reviewing details of corpora, editing metadata and deleting the corpora owned by any Tilde MT Service Subscriber, etc. The SA may also change SMT System definition scripts for any Tilde MT Service Subscriber/any SMT System.

Trained SMT Model means a non-downloadable set of files (language models, translation models, etc.) that are the result of the SMT system training process.

Object means corpora, language models, trained SMT Systems within the Tilde MT Database.

Resource Repository means a set of files (language models, translation models, etc.) that are the result of SMT system training process and can be used to run the Moses decoder translation processes.

Trained SMT System means set of models that are result of SMT system training process and it can be used to run Moses decoder translation processes.

Training Data means corpus (mono or multilingual) and other data used to train SMT System.

User means the Tilde MT Service Subscriber and/or Tilde MT User.

2. The Subject Matter

This Agreement regulates the rights and obligations of the Parties with regard to the provision of the Data, as defined in Section 1 hereof, to the Tilde MT Resource Repository and as well as sets forth the licensing terms and conditions for:

  • The use of the Data within the Tilde MT System
  • The Data accessibility to Tilde MT Service Subscribers and Tilde MT Users
3. License Grant

The Data Provider grants the Tilde MT Consortium the unrestricted, royalty free, worldwide license to use the Data for the purposes of the Project under the terms and conditions defined in this Agreement (hereinafter referred to as "License") by selecting one of the three types of licenses:

  • Public Data License: the Data are available for public use, i.e. to all Tilde MT Service Subscribers and Tilde MT Users that are allowed to train SMT Systems within the Tilde MT System (hereinafter referred to as "Public Data"). All Tilde MT TA or PU Users are allowed to look into the metadata, but not to the resources themselves, pick out resources for training data and carry out training tasks. The Tilde MT Consortium is at all times entitled to use the uploaded Public Data for training and for testing purposes.
  • Private Data License: the Data are available for private use only, i.e. only to the User who is also a Data Provider (hereinafter referred to as "Private Data") and its authorized Tilde MT Users. Other Tilde MT Service Subscribers and their respective Tilde MT Users shall have no access to the uploaded Private Data. This type of license is particularly appropriate for any data proprietary to the Data Provider, e.g. intellectual property rights protected or confidential data.
  • Shared Data License: the Data will be available only for certain user-defined groups, i.e. only to the Data Provider and specific Tilde MT User(s) with whom the Data Provider elected to share the Data under a separate sharing agreement concluded between the Data Provider, the Tilde MT Consortium and the respective Tilde MT User(s) and only for the purpose of training SMT Systems within the Tilde MT System (hereinafter referred to as "Shared Data"). The Shared Data shall be available only to the User who is also a Data Provider, absent separate sharing agreement.

The intellectual property rights of the Data uploaded to the Tilde MT System by the Data Provider will still belong to either the Data Provider or to the third party owner who had granted the Data Provider the right to upload and manage the Data. Neither Tilde MT Consortium nor any Tilde MT Service Subscribers or Tilde MT Users, except the Data Provider who owns the intellectual property rights to the Data or had been granted the right to modify it, shall be authorized to modify or alter any content of the Data.

The Data shall be processed in different ways by the Tilde MT System for the purposes of format conversion, quality assurance and their optimization for training needs. The Tilde MT Consortium, at its own discretion, is entitled to delete or otherwise remove any uploaded Data it deems of poor quality or infringing intellectual property rights or otherwise inappropriate.

4. The Data Provider's Obligations & Warranties

The Data Provider shall appoint a Tilde MT Service Subscriber Administrator (TA) authorized to upload the Data to the Resource Repository and administer the uploaded Data. As part of the upload process the TA shall follow the upload guidelines and specify information about the Data using the proposed metadata schema, apply the Data in the right formats and encoding and indicate the intellectual property and licensing conditions for the Data in accordance with the license granted under Section 4 hereof.

The Data Provider represents and warrants that the Data are of good quality, viruses, worms or other malware free, and their use hereunder shall not infringe third party intellectual property rights. The Data Provider does not warrant the correctness of the Data.

The Data Provider represents and warrants that the Data contain no proprietary or confidential information or personal data, unless the Private Data License will be applied to the Data.

The Data Provider is responsible for ensuring that the IPR of the Data are not infringed. The Data Provider shall only upload the Data either owned by the Data Provider or where the Data Provider has pertinent agreements with the intellectual property rights owner that covers the right to upload the Data to the Tilde MT System. Tilde MT consortium shall not be considered responsible for any infringement of IPR rights of the third party by Data Provider. The Tilde MT consortium makes no warranties as to the IPR of the publicly available Data.

In the event that the Data provided by the Data Provider causes injury to or violates third party intellectual property rights, the Tilde MT Consortium is entitled to claim full remuneration of the damage incurred in connection with such infringement, including any remuneration the Tilde MT Consortium would be ordered to pay to any third party intellectual property rights owner under a final and binding ruling by a competent court due to the infringement, as well as any pertinent legal costs.

5. Data Protection & Confidentiality

The safety of the Data uploaded to the Resource Repository shall be ensured by not allowing inspection or download of the Training Data or Trained SMT Models resulting from the training, except by the Tilde MT System Administrator (SA). The Tilde MT Users will not be able to get access to the Training Data in neither uploaded nor source format. The Training Data will be used only for training the Tilde MT System and the trained SMT models will only be available for usage through the Tilde MT System while all the temporary copies generated during the Training will be deleted immediately

Metadata for the Data will only be viewable and searchable for the Training Data that the respective User is allowed to use.

In case of intellectual property rights or Data confidentiality infringement or other unauthorized use of the Data, the Data Provider has the first right to bring action to prevent or terminate such infringement. If the Data Provider does not want to bring action, the Tilde MT Consortium may elect to take over and pursue with the consent of the Data Provider.

6. Term and Termination

This Agreement is concluded for indefinite term.

Either Party may cancel this Agreement at any time without any reason with a six week notice as of the last day of the month in which the cancellation notice was sent. The cancellation notice may be sent in writing, by facsimile or e-mail or any other appropriate means securing sufficient evidence of the cancellation and the date it was sent. The effect of such cancellation will be discontinuation of the Data upload and Tilde MT Services use hereunder, but it shall not affect the License regarding already uploaded Data, unless the Parties agree otherwise in writing.

Either Party may terminate this Agreement with immediate effect if the other Party is in material breach of its obligations hereunder.

6. Miscellaneous

This Agreement shall be governed and construed in accordance with the laws of your country.

In case of any dispute in connection with this Agreement, the Parties shall endeavor to resolve them amicably. If amicable resolution fails, the Parties agree to the venue of the competent court in your country.