Why Disambiguation Is the X-Factor for Your Text Analytics
The jaguar eats: what does this phrase mean? A disambiguation definition
The jaguar eats: what does this phrase mean? The answer is… it depends on the context. The reason is very clear if we consider these two sentences:
The jaguar eats meat.
The jaguar eats gas.
In the first sentence, we can understand that the jaguar in question is a panther, an animal that eats meat, the flesh of an animal. In the second sentence, the context changes completely with the word “gas”: Jaguar is a luxury car that consumes gasoline.
So, if we are to provide a disambiguation definition, we can describe it as the process of resolving conflicts that arise when a term can express (like “jaguar” and “eat” in the previous sentences) more than one meaning, leading to different interpretations of the same string of text.
The first difficulty in disambiguation is that language contains many ambiguities that must be resolved in order to identify the correct meaning of words and sentences based on different contexts. The second difficulty is that the number of existing combinations of terms increases exponentially since language is always evolving.
When it comes to text analytics, it’s easy to understand why disambiguation is the crucial problem that must be resolved.
Human vs. machine
For a human, disambiguation is not a real problem. The meaning is obvious because of our ability to automatically make several calculations based on our intellect, memory, culture, context, education, experience that help us understand the meaning of a word. As a result, we can disambiguate both spoken and written language at various levels of difficulty based on what we have learned.
On the contrary, a machine can’t deduce meaning automatically because it doesn’t have a “natural” reference system that supports it in this process. However, a technology must somehow be able to “reason,” otherwise it couldn’t support text analytics activities, so how is this possible? Thanks to artificial intelligence.
The machine must be “taught,” but it can also learn from experience with the proper training! It needs a system that provides it access to background information, in the way that experience or education serves humans, that allows it to cope with ambiguities in text in a way that it can determine the proper meaning. Such a system, combined with a multi-level text analysis and a knowledge graph that contains millions of concepts, results in a powerful module capable of logical and comprehensive text understanding on a large scale.
The disambiguation process
At the core of Expert System’s Cogito cognitive platform is a sophisticated disambiguator that is able to solve ambiguities and understand the meaning of each word in context. This is possible thanks to a multi-level text analysis that consists of different, consecutive phases (lexical, grammatical, syntactical and semantic analysis) that lead to a cognitive and conceptual map of texts, which constitutes the final output of the disambiguation process.
Working with “concepts,” rather than simple keywords, Cogito can fully understand the meaning of single words, sentences and entire documents and allows users to reach accuracy levels over 90% in totally open contexts. Moreover, thanks to the latest AI enhancements, Cogito has improved its text analytics capabilities to make it even more effective in disambiguating text acruss multiple languages.
This is Cogito. What are you waiting for?