Presentations

Abstract

In the beginning of 2018, The Language Archive migrated its repository from an in-house built solution to a solution that is largely based on Islandora. We will talk about the migration trajectory and will present the new setup, which includes a custom ingest front- and back-end

Abstract

During the past 10 years, the Max Planck Institute for Psycholinguistics has developed an extensive technological framework around its digital archive for linguistic resources. About two years ago the MPI started installing archives based on this "Language Archiving Technology” (LAT) framework in various locations around the world. The idea behind this initiative is to have regional archives in the proximity of the area where the linguistic resources are collected. This will facilitate access to the resources and create more local involvement and awareness towards the preservation of endangered languages and cultures. The user interfaces of some of the LAT tools are not always very suitable for the speech community due to the language that is being used (English) and the extensive set of features of these tools, many of which are of less interest to the speech community. Therefore a framework was developed that allows the integration of archived content within a web portal that is managed using a standard Content Management System (Plone). A web service was developed that enables searching of the archive’s metadata database using the SOAP protocol. From within the CMS, the content editor can easily specify queries for specific metadata values, e.g. all songs in a particular language. These queries can be linked to buttons or images in the portal. The search results are then parsed into nicely formatted lists of resources. The facility will make the use of the local archive more efficient and user friendly. www.lat-mpi.eu www.mpi.nl/dobes www.plone.org

Abstract

Technological innovation is continuously creating new encoding formats for video. The introduction of HDTV, the wish to move towards 3D video etc will increase the required bandwidths and capacities by factors. New coding standards such as H.264 and JPEG2000 have been developed to overcome the problem of increasing bit rates and new codecs such as H.265 are in the pipeline. In addition we have seen in the recent decades that the maintenance of old formats is not guaranteed if their markets become too small.
This extreme innovation rate is problematic for all archiving intentions, since archiving means guaranteeing continuous accessibility of the archived digital resources. It is known that a continuous migration will be required to interpret stored video streams. At the bit-stream level migration to new storage technology can be organized by fully automatic procedures. At the encoding level problems are much more severe. When migrating compressed video for example we will be confronted with concatenation effects creating serious artifacts. Ideally we would like to store uncompressed or lossless compressed video so that we have a master copy from which we can generate the various presentation formats. Currently, frequently MPEG2 is used for this purpose although it does not prevent information degradation due to concatenation. We will argue for a move to lossless JPEG-2000 encoding as master format and proper process metadata description. Yet we have to solve the dilemma that field workers will deliver highly compressed formats due to the usage of consumer equipment also in future.

Abstract

The repository of the MPI contains different types of linguistic material: the DOBES endangered languages archive, the ESF second learner corpus, the Dutch Spoken National Corpus, MPI's gesture corpora, MPI acquisition corpora and MPI language documentations of the language and cognition research group. The archive covers more than 200.000 objects, mostly organized in sessions that are described with the IMDI-based metadata descriptions. Mostly, these sessions contain digitized audio/video signals and layers of annotations. In general access to these resources is limited and can be made available upon request. The Language Archiving Technology (LAT) is meant to contribute to the archive infrastructure. It focuses on open accessibility of the language resources; it supports dynamic and continuously enriched collections according to the Live Archives ideas; it stresses the need for long-term archiving of our digital collections covering unique martial about languages that will probably be extinct in a few decades and it follows the trend towards service oriented architectures. LAT components consist of data management and ingestion tools (IMDI, LAMUS and AMS) and of archive enrichment and visualization tools (ELAN, ANNEX and LEXUS). The tools are being developed and maintained by the Technical Group of the MPI. All LAT products are or will become available under an Open Source license and will be available free-of-charge in academic research.