NeurotimeGuide
How MediaEye's AI handles relevance, sentiment, categorization and grouping
Dashboards show tidy charts, but each number depends on four AI decisions. Understanding them shows why the results can be trusted.
Relevance: is it about us?
A keyword match is not a mention. A keyword like “Bakı Kart” could match a transport-card news item, a bank product and a forum post comparing both. MediaEye's relevance models keep only what is about the tracked subject. Value: your team reads signal, not noise, and every chart is built on real mentions.
Sentiment: what is the tone?
Each mention is classified as positive, neutral or negative toward the subject. A report saying “the bank denied the allegations” is usually negative coverage even though the statement is defensive. Value: you see at a glance whether coverage is helping or hurting, and alerts can trigger only on negative mentions.
Categorization: what is it about?
Mentions are assigned to topics — product, service quality, pricing, leadership, CSR — tailored to your business questions. MediaEye categorizes both text and visual content, using OCR for text inside images. Value: you know not just how much you were mentioned, but why.
Grouping: is it the same story?
Reposts of one press release and different articles about the same event are grouped. Counting 40 reposts as 40 stories exaggerates; collapsing everything hides how far a story spread. Value: you see real reach of a story and the outlets that carried it.
More detail: methodology.