Skip to main content

Semantic mapping is the process of creating a visual representation of related concepts using a web or word cloud. This process helps humans understand and retain new concepts and the words to describe them by linking the new information to existing knowledge. Studied since the 1980s, semantic mapping is a valuable technique for increasing vocabulary and therefore reading comprehension for “students who are learning disabled, students in remedial reading classes, and illiterate adults, as well as those in regular classrooms.”  

Just a few years later, researchers formalized this approach to related concepts with “self-organizing semantic maps.” A set of training data noted whether or not an animal possesses a specific characteristic, such as the ability to fly, or if has two or four legs. When this dataset is visualized, the resulting map indicates relevance, or the degree to which each animal in the dataset is semantically similar. The authors of this study argue that this neural network (while simplified) may replicate the way we understand relationships between concepts. 

Historically, artificial intelligence has operated in a “black box,” limiting the transparency around its decision-making. Formalized semantic mapping gives humans and AI a way to communicate using concepts both parties understand. 

Semantic Mapping in Patent Searching

IP.com puts the power of semantic mapping to use with its AI-enabled patent search software. InnovationQ Plus® uses concept-based search technology to identify the most relevant results within the global patent database. The results can then be visualized with our Semantic Map, which indicates the conceptual relationships between patents. 

Without semantic mapping, this geographic representation of concepts would not be possible. Not only does InnovationQ Plus® form a word cloud of related ideas, it then uses natural language processing to map individual patents to these concepts based on their claims. This powerful patent mapping technique allows companies to identify related spaces that are highly competitive, home to emerging markets, or void of patents altogether.