Temnos Tags & Topics

tags2Temnos Tags and Topics are dynamically-generated metadata about concepts that are discovered in documents. Unlike typical entity-extraction systems, Temnos Tags and Topics do not require predefined vocabularies (word lists) and are not limited to proper names (e.g., San Francisco, Barack Obama, Apple Inc.).

Temnos Tags consist of the concepts that are pertinent to a document, whether that concept is a single word, a collocation of two or more words, or a proper name. For example, an article about the SF Giants winning the world series might be assigned the tags “San Francisco Giants”, “baseball”, “home run”, “World Series”, “four-game sweep” and “Sergio Romo” because they were important concepts in the news story.

All Temnos Tags are accompanied by a relevance score (a decimal value between 0.0 and 1.0, inclusive) that indicates how strongly the term represents the document to which is it associated. High-scoring tags are the best candidates for use in tag clouds, linking to related stories, and creating very specific collections of related content.

Temnos Tags vs. Keywords

Although keywords are commonly used for searching and targeting content, they are inherently problematic by nature. Most words in the English language have more than one meaning (i.e., they are polysemous). For example, “mercury” can refer to an element, a planet, an automobile brand, a NASA space program, a Roman god, an astronomy magazine, and more. Even a common word such as “orange” could refer to a color, a fruit, a band, a telecommunications company, a film, or a city.

As a result of polysemy, accurately expressing the correct meaning of a keyword requires the use of yet more polysemous keywords and, if a system supports it, keywords that match what is not desired (i.e., negative keywords). However, the skill and labor involved in researching, discovering, and assembling the right collections of keywords for different uses is expensive and inefficient.

Temnos Tags, on the other hand, are generated with their senses differentiated. A tag that has five senses, for example, is represented by five different tag identifiers so that the correct meaning can be easily selected.

Temnos Topics

Temnos-metadata-comparison
Figure 1: Comparison of four types of Temnos metadata

The specificity of Temnos Tags is not always needed to effectively organize or identify content. Often, higher-level concepts are sufficient or simply more desirable. Temnos Topics represent broader themes that are present in document collections.

As illustrated under the Technology category in Figure 1, specific Temnos Tags such as the brand names of portable hard drives that are mentioned in a consumer electronics product review would be represented by the Temnos Topic, “Portable Storage”. Thus, Temnos Topics are roll-ups of related Temnos Tags and are often used for creating topic pages, topic indexes, and cross-pollinating traffic to related content between sites.

For higher-level metadata, see Temnos Categories and Temnos MetaTopics.

Contact us to get started!