Skip to main content

Enhancing and Re-Purposing TV Content for Trans-Vector Engagement

Deliverables

Requirements for Content Owner Use Case

D5.1: Requirements for Content Owner Use Case (NISV: M10). This document will specify the requirements regarding data visualizations and reporting, engaging selected experts from RBB and ZATTOO to inform WPs 1-4. In particular, this document will be built by performing an inner cycle of focus group and collaborative requirements elicitation sessions on top of which some selected experts will provide feedbacks and fine tuning.

Dissemination Report

D7.3: Dissemination Report (Genistat/ZATTOO: M6). This document is a quantitative and qualitative report on planned and completed communication and dissemination activities (including the stakeholder forum) and measures their success against the stated goals - recommending modifications to the original plans where necessary. It will be updated every 6 months internally.

Requirements for Consumer Use Case

D6.1: Requirements for Consumer Use Case (RBB: M10). Specifies requirements of consumers to inform WP1, WP2 and WP3 on required components and functionalities.

First validation of Engagement Monitoring Prototype

D5.2: First validation of Engagement Monitoring Prototype (NISV: M20). The validation results of the use case will comprise qualitative and quantitative data, and include a comparative analysis of user interfaces and the applicability in daily use by content owners and media professionals, focused on qualitative analysis gathered from the selected professionals of the project partners who will be involved in the evaluation activities.

Second validation of Engagement Monitoring Prototype

D5.3: Second validation of Engagement Monitoring Prototype (Genistat: M39). The validation results of the use case will comprise qualitative and quantitative data, and include a comparative analysis of user interfaces and the applicability in daily use by media professionals, reporting on extended use among a wider audience of professionals, gathering their feedback and suggesting final TVP improvements.

First validation of Personalization Prototype

D6.2: First validation of Personalization Prototype (Genistat/ZATTOO: M20). This deliverable consists (1) of the first prototypes for the evaluation, an initial adaptation of existing consumer applications incorporating the newly developed TVP features and functionality, especially Content Scheduling and Content Recommendation, and (2) a report on the results of the first longitudinal user tests which informs WPs 1-3 on required frontend and backend improvements and will deliver updated insights on acceptance, including usability and ethical issues.

Second validation of Personalization Prototype

D6.3: Second validation of Personalization Prototype (RBB: M39). This deliverable consists of updated prototypes for evaluating and demonstrating the TVP functionality and a report on the evaluation results. The report will focus on overall project results with respect to customer acceptance and provide guidance for future application of the TVP.

Data Ingestion, Analysis and Annotation, Second Version

D1.2: Data Ingestion, Analysis and Annotation, Second Version (CERTH: M20). Describes initial set of data ingestion services delivered to TVP, covering partner CMS/MAMS, social media, Web logs, video/TV viewing logs. Includes further developments and evaluation of improvements for video abstractions. Presents the updated and extended versions of concept-based video abstractions. Specifically, MTL and structured outputs will be introduced in our deep learning architectures targeting more accurate concept detection, improved fine-tuning algorithms will be developed that consider the commonalities between the source and target domain, the developed algorithms will be scaled to a bigger pool of target concepts. Includes further developments and evaluation of improvements for brand detection. Specifically, we will extend the brand detection algorithms with inductive transfer in our CNN architectures, more brand labels and advanced fusion methods for the final video-level brand detection.

Metrics-based Success Factors and Predictive Analytics,final version

D2.3: Metrics-based Success Factors and Predictive Analytics, final version (MOD: M30) This document presents the final version of the success factors and predictive analytics based on multiple emotional categories and measures of disagreement calculated for arbitrary time intervals and with an on-the-fly reconfiguration of weights, pattern detection from near real time audience and viewer data metrics, and a hybrid prediction model based on most accurately evaluated analytics for each combination of vector and metric. The combined model is retrained dynamically depending on the amount of new data.

Content Adaptation, Re-Purposing and Scheduling,first version

D3.2: Content Adaptation, Re-Purposing and Scheduling, first version (VUA: M20). This document will report on an initial development and evaluation of the video adaptation, re-purposing and scheduling technologies as well as on content recommendations on a best effort basis, using the first version of the predictive analytics (output of the T2.4) to generate recommendations according to audience segment, based on a final version of viewer profiling and the shared metadata model.

Trans-Vector Platform, TVP Dashboard and Revised

D4.2: Trans-Vector Platform, TVP Dashboard and Revised Prototype (WLT: M24) delivers a revised version of the platform including API framework and TVP Dashboard (T4.3), reflecting user feedback from the uses cases (WP5, WP6) in terms of design and usability. The customizable dashboard will include the T4.2 visualization and support cross-lingual exploration and visualization of content streams across vectors, languages (English, French, German), and other context dimensions.

Metrics-based Success Factors and Predictive Analytics,first version

D2.2: Metrics-based Success Factors and Predictive Analytics, first version (WLT: M20) This document presents the first version of the success factors and predictive analytics based on the derived annotations and metrics, covering: placement of content parallel to future events (T2.1); sentiment detection and desired and undesired associations calculated as a weighted 24 hour moving average (T2.2); trend detection from historical audience and viewer data (T2.3); and a generic prediction model independent of vector (T2.4).

Data Ingestion, Analysis and Annotation, First Version

D1.1: Data Ingestion, Analysis and Annotation, First Version (CERTH: M10). Outlines identified media vectors to support in TVP and implementation plan for each based on existing APIs and requirements. Describes initial deployment of concept-based video abstraction technologies with benchmarking evaluation. Describes initial deployment of brand detection solution with benchmarking evaluation. Reports on the chosen metadata solution and set-up, outlines first deployment of metadata mappings and entity extraction services to annotate ingested content according to the model and vocabulary and storage of annotations as a Knowledge Graph without performance optimizations.

Temporal Annotation and Metrics Extraction

D2.1: Temporal Annotation and Metrics Extraction (MOD: M10) This document presents the plans for implementing the event extraction capabilities on top of MOD’s entity extraction tool Recognyze and the temporal annotation of content items. It also identifies the content-based and audience metrics to be measured and outlines the intended approach and implementation to achieve the metric extraction across published vectors.

Content Adaptation, Re-Purposing and Scheduling,final version

D3.3: Content Adaptation, Re-Purposing and Scheduling, final version (GENISTAT: M34). This document will report on the final video adaptation and re-purposing technologies - more accurate video skimming by using high-light detection in combination with the visual concept detection results and more discriminative text-to-vector representations that will be combined with improved CNN architectures for hot-spot detection - as well as scheduling and recommendation technologies - using the final version of predictive analytics to target content on vectors at the individual viewer level.

Data Analysis and Annotation, Final Version

D1.3: Data Analysis and Annotation, Final Version (CERTH: M30). Presents the final set of ReTV data analysis and annotation techniques. The final version of concept-based video abstraction and brand detection technologies. Updates to support final model and vocabulary aligned to WPs 2 and 3, support all media vectors, and Knowledge Graph optimizations at data scale based on queries/views needed by modelling and analytics processes (WP3).

Trans-Vector Platform, Final and Optimized Version

D4.3: Trans-Vector Platform, Final and Optimized Version (WLT: M36) will deliver an optimized platform in terms of response times and scalability (T4.4), including revised versions of all the embedded visual tools, API specifications, and interactive controls for drill down and on-the-fly query refinement.

Metadata and Viewer Profiling

D3.1 Metadata and Viewer Profiling (VUA: M10). This document presents a first version of a metadata model & vocabulary based on aggregation of required properties and value spaces across all data produced in ReTV, mappings to this shared model and a viewer profile model expressed according to this model.

Website and online communication

D7.1: Website and Online Communication (NISV: M3). A project website and social media channels will be set up and maintained throughout the project for communication activities.

Searching for OpenAIRE data...

Publications

VERGE in VBS 2021

Author(s): Stelios Andreadis, Anastasia Moumtzidou, Konstantinos Gkountakos, Nick Pantelidis, Konstantinos Apostolidis, Damianos Galanopoulos, Ilias Gialampoukidis, Stefanos Vrochidis, Vasileios Mezaris, Ioannis Kompatsiaris
Published in: MultiMedia Modeling - 27th International Conference, MMM 2021, Prague, Czech Republic, June 22–24, 2021, Proceedings, Part II, 12573, 2021, Page(s) 398-404, ISBN 978-3-030-67834-0
Publisher: Springer International Publishing
DOI: 10.1007/978-3-030-67835-7_35

Online News Monitoring for Enhanced Reuse of Audiovisual Archives

Author(s): Rasa Bocyte, Johan Oomen, Lyndon Nixon, Arno Scharl
Published in: Digital Libraries for Open Knowledge - 24th International Conference on Theory and Practice of Digital Libraries, TPDL 2020, Lyon, France, August 25–27, 2020, Proceedings, 12246, 2020, Page(s) 243-248, ISBN 978-3-030-54955-8
Publisher: Springer International Publishing
DOI: 10.1007/978-3-030-54956-5_18

Content Adaptation, Personalisation and Fine-grained Retrieval: Applying AI to Support Engagement with and Reuse of Archival Content at Scale

Author(s): Rasa Bocyte, Johan Oomen
Published in: Proceedings of the 12th International Conference on Agents and Artificial Intelligence, 2020, Page(s) 506-511, ISBN 978-989-758-395-7
Publisher: SCITEPRESS - Science and Technology Publications
DOI: 10.5220/0009188505060511

AI4TV 2020 - 2nd International Workshop on AI for Smart TV Content Production, Access and Delivery

Author(s): Raphaël Troncy, Jorma Laaksonen, Hamed R. Tavakoli, Lyndon Nixon, Vasileios Mezaris, Mohammad Hosseini
Published in: Proceedings of the 28th ACM International Conference on Multimedia, 2020, Page(s) 4756-4757, ISBN 9781450379885
Publisher: ACM
DOI: 10.1145/3394171.3421894

Video Analysis for Interactive Story Creation - The Sandmännchen Showcase

Author(s): Miggi Zwicklbauer, Willy Lamm, Martin Gordon, Konstantinos Apostolidis, Basil Philipp, Vasileios Mezaris
Published in: Proceedings of the 2nd International Workshop on AI for Smart TV Content Production, Access and Delivery, 2020, Page(s) 17-24, ISBN 9781450381468
Publisher: ACM
DOI: 10.1145/3422839.3423061

Predicting Your Future Audience's Popular Topics to Optimize TV Content Marketing Success

Author(s): Lyndon Nixon
Published in: Proceedings of the 2nd International Workshop on AI for Smart TV Content Production, Access and Delivery, 2020, Page(s) 5-10, ISBN 9781450381468
Publisher: ACM
DOI: 10.1145/3422839.3423062

DataTV 2019: 1st International Workshop on Data-Driven Personalisation of Television

Author(s): Jeremy Foss, Ben Shirley, Benedita Malheiro, Sara Kepplinger, Lyndon Nixon, Basil Philipp, Vasilieos Mezaris, Alexandre Ulisses
Published in: Proceedings of the 2019 ACM International Conference on Interactive Experiences for TV and Online Video - TVX '19, 2019, Page(s) 286-292, ISBN 9781-450360173
Publisher: ACM Press
DOI: 10.1145/3317697.3323349

The Trans-Vector Platform for optimised Re-purposing and Republication of TV Content

Author(s): Lyndon Nixon, Miggi Zwicklbauer, Lizzy Komen, Basil Philipp
Published in: 2019
Publisher: DataTV 2019: 1st International Workshop on Data Driven Personalisation of Television

AI for Audience Prediction and Profiling to Power Innovative TV Content Recommendation Services

Author(s): Lyndon Nixon, Krzysztof Ciesielski and Basil Philipp
Published in: 2019
Publisher: ACM Multimedia 2019, October 2019, Nice, France

Improving Named Entity Linking Corpora Quality

Author(s): Weichselbraun, A., Brasoveanu, A., Kuntschik, P. and Nixon, L.
Published in: 2019
Publisher: RANLP 2019, Varna, Bulgaria , September 2019

Automatically Adapting and Publishing TV Content for Increased Effectiveness and Efficiency

Author(s): Basil Philipp, Krzysztof Ciesielski and Lyndon Nixon
Published in: 2019
Publisher: ACM Multimedia 2019, October 2019, Nice, France

A Stepwise, Label-based Approach for Improving the Adversarial Training in Unsupervised Video Summarization

Author(s): Evlampios Apostolidis, Alexandros I. Metsai, Eleni Adamantidou, Vasileios Mezaris, and Ioannis Patras
Published in: 2019
Publisher: ACM Multimedia 2019, October 2019, Nice, France

In Media Res: A Corpus for Evaluating Named Entity Linking with Creative Works

Author(s): Adrian M.P. Brasoveanu, Albert Weichselbraun, Lyndon Nixon
Published in: Proceedings of the 24th Conference on Computational Natural Language Learning, 2020, Page(s) 355-364
Publisher: Association for Computational Linguistics
DOI: 10.18653/v1/2020.conll-1.28

Performance over Random - A Robust Evaluation Protocol for Video Summarization Methods

Author(s): Evlampios Apostolidis, Eleni Adamantidou, Alexandros I. Metsai, Vasileios Mezaris, Ioannis Patras
Published in: Proceedings of the 28th ACM International Conference on Multimedia, 2020, Page(s) 1056-1064, ISBN 9781450379885
Publisher: ACM
DOI: 10.1145/3394171.3413632

Video Summarization Using Deep Neural Networks: A Survey

Author(s): Evlampios Apostolidis, Eleni Adamantidou, Alexandros I. Metsai, Vasileios Mezaris, Ioannis Patras
Published in: 2021
Publisher: ARXiV

Topics Compass: uncovering Trending Topics for Optimised Media Content Publication

Author(s): Nixon, L., Scharl, A. and Bocyte, R.
Published in: 2020
Publisher: ACM IMX 2020