In this post Romina Gersuni, Global Strategy Manager at pressrelations GmbH, shares her views on human and machine intelligence and on how Artificial Intelligence is affecting media monitoring. She talks about the main advantages and disadvantages of programmatic technology as well as the Multisensor project and explains how it affects our industry.
Question: The pairing of human and machine intelligence! What is your experience with Artificial Intelligence and how does it affect media monitoring?
Romina Gersuni: pressrelations believes in a hybrid approach. Therefore, our AI technologies serve as a support for the human analyst. We have integrated machine intelligence into our back-end solution PR-Office, for example. PR-Office supports the editorial teams along the management, validation and categorization of data, and this process can be done in an automatic, semi-automatic or manual workflow. The human analyst is responsible for a highly qualitative analysis and results.
In a concrete sense, pressrelations uses AI for automatic content extraction of text documents. Here, different methods of Natural Language Processing (NLP) are used for:
- Article selection
- Topic clustering
- Article categorization
- Plagiarism detection
- Keyword extraction
- Named entity recognition (NER)
- Quote extraction
- Sentiment analysis
Thus, AI is already affecting our industry helping us to work in an increasingly efficient manner, reducing production times, enabling more attention to details and granting all teams a chance to renew their efforts towards the addition of real value to the data.
Programmatic technology scales through automation and it is getting smarter and better over time as it collects more data. What do you believe are the main advantages and disadvantages of programmatic technology?
The media intelligence industry has been in a race towards automated or programmatic solutions, and indeed this kind of solutions can make processes more effective, as long as the right data is applied.
Some of the positive effects of programmatic technologies include the chance of narrowing the target data accordingly and learning from the client usage. This technologies include behavioral targeting capabilities (through cookies), and thereof they are extremely useful once applied to the clients’ front-end solution (e.g., Google Analytics).
At the same time in terms of the behavioral targeting capabilitiesthe cross-device efficiency of programmatic technologies is challenged. And if your company is focusing on ad targeting, fraud is still a big problem (URL masking), therefore a fraud monitoring program should be included in such systems.
pressrelations participates in the Multisensor EU funded research project. Can you please explain us what this project is about, what is your role in this project and how will it affect media monitoring?
MULTISENSOR aims at advancing the research and development of multilingual media analysis technologies for the support of media monitoring companies, SMEs and journalists. The project started in 11/2013 and is running for 3 years, carried out by a consortium of nine international partners such as EURECAT, Ontotext, and Linguatec.
Within this project seven major technological fields are being developed: Automated translation, speech-to-text conversion, summarization, topic clustering, contextual information, sentiment analysis and semantic search options.
pressrelations has a multifaceted role in the project and is in charge of use case design, system development and evaluation. Furthermore, pressrelations will lead the exploitation activities.
Neueste Artikel von Romina Gersuni (alle ansehen)
- “Watch out for developers out of the cage!” FIBEP TechDay 2019 - 29. Mai 2019
- The Inside Story: about the FIBEP Journey - 17. September 2018
- AMEC summit Barcelona 2018 – Interview with Rayna de Lange - 5. Juli 2018