As part of MICO enterprise use-cases, we are currently implementing a Enterprise Cross Media Search solution, integrating MICO with Sensefy an open source enterprise federated search framework. The goal of this project is to utilize the power of MICO cross media analysis features to boost enterprise search with semantic search across all media file types. This blog post gives a progress update of our use-case development project.

MICO enterprise cross media search use-case

Sensefy MICO integration aims to cover following use-cases;

  • The search user can search across heterogeneous content repositories (file-system, Alfresco, Box etc)
  • The search user can search across all media files (text, images, audio, video)
  • The search user is given keyword suggestions and entity/concept suggestions as auto completion of the search queries typed
  • When selecting an entity or concept from the suggestions, all media files which mention about that particular entity  or concept should be included in the search results

MICO configuration

As enterprise content could be of any media type (text, image, audio, video), we need to extract metadata from all those files, to enable semantic search to search by entities and concepts.

Following is the extractors pipeline we have configured in the MICO server for the enterprise cross media search use-case. We aim to extract metadata from text, images, audio and video files by orchestrating the available MICO extractors in the following configuration.

extractors pipeline

MICO extractors pipeline for enterprise cross media search

MICO Connector

We are currently developing the MICO transformation connector for Apache ManifoldCF which is used as the document ingestion engine in the enterprise search solution.  This connector will inject the content to MICO broker endpoint to extract metadata and store the results in the MICO Marmotta triple store.

MICO Java client

One of the goals of MICO exploitation is to make MICO technology easily adaptable by a wider audience. This includes providing  sample codes, tools and documentation on MICO service utilization. In this aspect, we are developing a general purpose Java client for MICO to use MICO services. The client encapsulates MICO metadata model and provides Helper classes to perform content injection and metadata querying. You can find the mico-client project on github here. (Please note that it’s currently under development)

Stay tuned for more updates on the MICO enterprise cross media search use-case development…