Atlastic.ai offers clients an unmatched pipeline of high quality, noise-reduced, structured and enriched media data from millions of sources to integrate directly into the preferred IT infrastructure and platforms. Our unique combination of 6 advanced AI systems, working together to provide the most comprehensive and informative media data set, enables organizations to better understand the full relevant context and make better-informed decisions. Here are some more details on our technology.
Crawling is not just crawling. Atlastic.ai’s fully automatized and optimized crawling is automatically adapting to any source’s specific construction without manual intervention. Due to this automation, our coverage is the most comprehensive in the industry. We detect navigation links, recognize already crawled links and resynthesize an optimal crawling schedule after each crawling session. Using browser simulation, combined with big data and low-level network engineering, we automatically capture a visual of a web page in real-time. To ensure that our coverage stays the most comprehensive, we scan newly registered domains and perform a semantic similarity analysis to determine if its contents is similar to the domains already in our media monitoring database.
In a world of endless media information, we are combining several strong AI’s to make the understanding a lot easier and faster for the user. By accurately determining the publication date, our media data enables identifying media patterns spread out over time. We interpret the right time of publication for any language and for any cultural convention. This happens by first identifying the text indicating the publication date, followed by a semantical analysis over that text. Our advertising detector removes the ads that add noise to monitored earned media. This by using semantic similarity analysis, to calculate the distance of the text being either an ad or normal article text. Print media is analyzed using computer vision, natural language processing and automated reasoning techniques to transform a stream of pages into a stream articles. By using speech to text algorithms, we transcribe the spoken text in radio shows and TV programs to text.
The reputation analysis, combined with time-series analysis, is fundamental in identifying and tracking reputation trends. Deep-learning based natural language processing is used to classify a given text to either negative, neutral or positive. Image quality is scored and ranked as well.
The origin of any information needs to be clarified. Our statistical algorithms automatically identify the language of a mention. Audience data is combined with language data to estimate the main market the article has the most impact on.
Connection the data points is a crucial part of crystalzing the insights for the user. Atlastic’s deep learning NLP is identifying named entities like products, brands, companies, organizations, persons and countries that are of interest to be monitored. Media mentions that are about the same topic, but occurring on different media outlets are grouped together in the same story cluster and – with our Salience API component – are we differentiating between the very important entities from the less important ones.
The audience for the media mentions is a fundamental metric to compare the media impact of one to another. We combine best-in-class audience data with our in-house crawling metrics to estimate the likely number of people that have seen this article, the impressions. The AI techniques regression, classification and time Series Analysis are used here. Deep-learning based NLP is used to determine the category and type of a mention. All of this to give the users a better understanding of the character of the mention.