Lyriko Similarity
Lyriko Similarity leverages AI to identify relationships between documents, facilitating content comparison and efficient tracking of text and image reuse (from global to local, or local to local).
This enables various use cases, including new content creation, reuse, and personalization.
Key Features
Document Similarity Detection
Document Comparison
Image Similarity Tracking
Content Relationship Mapping
Cross-Market Reuse Insights
How does it work?
- Compare: Lyriko analyzes documents and images to detect similarities across content.
- Identify Relationship: Lyriko identifies “Parent” and “Child” documents, tracking hierarchies across all the DAM.
- Identify Patterns: Lyriko provides insights on content reuse and duplication across local markets.
- Streamline Insights: Marketing Teams gain actionable data to enhance content adoption and optimize resource investments.
Expected Benefits
Maximize content reuse across markets
Detect duplicate documents and images
Optimize resources for efficient marketing
Cut costs for content creation
Speed up content delivery
FAQs​
What is content similarity?
Content similarity is the measure of how closely two pieces of content, documents or images, match in meaning, wording, and structure. It is used to detect duplicates, track content reuse, and map relationships between assets, so teams can avoid recreating material that already exists and optimize how content is produced.
What is Lyriko Similarity?
Lyriko Similarity is an AI-powered tool that identifies relationships between documents and images, making content comparison automatic and tracking text and image reuse across markets (global-to-local or local-to-local). It helps marketing teams maximize reuse, detect duplicates, optimize resources, and speed up content delivery.
Why do companies need content similarity analysis?
Large organizations produce and adapt huge volumes of content across markets, often recreating material that already exists. Without similarity analysis it is impossible to know what is being reused, duplicated, or ignored. Lyriko Similarity makes reuse visible, reduces duplicate work, optimizes budget, and improves the quality and consistency of produced content.
How does Lyriko Similarity work?
Lyriko works in four steps: (1) it compares documents and images to detect similarities; (2) it identifies “parent” and “child” relationships, tracking content hierarchies across the DAM; (3) it surfaces patterns of reuse and duplication across local markets; (4) it turns these into actionable insights to improve content adoption and optimize resource investment.
How does Lyriko calculate a similarity score?
Lyriko combines three signals into a final score: semantic similarity (meaning), lexical similarity (wording), and claims/message similarity (the underlying message). Each page is matched to its best counterpart, scores are averaged across pages, and the channels are combined with customizable weights (e.g. 40/40/20) to produce one overall similarity score per document pair.
What are the three similarity dimensions Lyriko measures?
Lyriko measures three dimensions. Semantic similarity captures overall meaning, so two pages covering the same topic in different words still score high. Lexical similarity measures literal text overlap, shared words and phrases, and is best for spotting near-duplicates and localized versions. Claim similarity compares the medical claims extracted from each document, which is critical for ensuring regulatory consistency of statements across materials.
Does content similarity work across different languages?
Yes, cross-language similarity is supported. The semantic dimension uses embeddings that capture meaning regardless of language, so documents in different languages covering the same topic still score high semantically. The lexical dimension will be low across languages, since it measures literal text overlap. Lyriko also supports language-pair weights, so you can weight semantic higher for cross-language pairs (e.g. EN–IT) and lexical higher for same-language pairs (e.g. EN–EN).
What is the global–local relationship matrix?
For each global–local document pair, Lyriko generates a matrix showing the overall similarity score for every page combination and highlighting pages with no measurable similarity. High scores along the diagonal indicate the local document follows the same communication flow as the global one; a sparse matrix indicates selective reuse, where pages were adapted, cherry-picked, or reordered.
Can Lyriko detect similarity between images, not just text?
Yes. Lyriko Similarity detects both document and image similarity. This lets teams track visual reuse alongside text, identify duplicate or near-duplicate images across markets, and understand which visual assets are reused most or least.
What are the main use cases for content similarity?
Clients typically use Lyriko Similarity for three goals: tracking content reuse across local markets, detecting duplicate documents and images, and supporting new content creation, reuse, and personalization. The calculation methods can be customized depending on which objective matters most.
How does Lyriko track content reuse across markets?
Lyriko compares global “parent” documents with local “child” versions, scoring how much each local asset reuses, adapts, or diverges from the source. This reveals reuse patterns across markets, shows which content is adopted or ignored, and helps headquarters and local teams align content strategy and budget.
Is there human oversight in the similarity analysis?
Yes. Lyriko uses automated matching with an optional “human-in-the-loop” double-check. The AI performs the comparison at scale, while authorized users can review and validate the matches, ensuring accurate verification of content consistency before insights are acted on.
What business value does Lyriko Similarity deliver?
Lyriko Similarity maximizes content reuse across markets, detects duplicate documents and images, optimizes resources, cuts content-creation costs, and speeds up content delivery. By revealing which specific elements are most or least reused, it also improves the quality of newly produced content and reduces time to market.
How is content similarity different from content tagging?
Content tagging assigns descriptive labels to individual assets so they can be found and organized; content similarity compares assets to each other to measure how alike they are and track reuse. Tagging answers “what is this content about?”, while similarity answers “how does this content relate to other content?”. In Lyriko the two modules complement each other.
What types of content and reuse relationships can Lyriko analyze?
Lyriko analyzes both documents and images, and maps reuse in multiple directions: global-to-local (how local markets adapt headquarters content) and local-to-local (how markets reuse each other’s content). It identifies parent–child hierarchies across the whole DAM and highlights both strong page-to-page alignment and selective or fragmented reuse.
Discover the other Lyriko modules
