Analysis

Impressions received by means of intuition are a primary data type in intuizen. An impression can be as simple as a single word, like the colour red, or it can be as complex as a scene or metaphor.

Reasoning

Increasing the throughput of intuition analysis is critical to scaling. By increasing throughput, both in terms of time, effort and quality level, we can help to increase the number of sessions that can be analyzed in a given time period. This analysis would be performed cheaply (relative to a human worker) by state of the art artificial intelligence, at a consistency and quality level that is manageable and optimistically will improve over time.

The problem of inter-rater reliability has previously been identifed in previous research source.

Impression data format

Raw impression data is organized as part of a Session transcript. Session transcripts can be uploaded manually by the user in the dashboard or captured with the help of the intuizen session monitor.

Methodology

Descriptor Processing

Each impression passes through a pipeline of processing. The end goal is to decompose the impression into a set of discrete descriptors that can be used to compare and contrast with the target. This step is akin to "refining" the raw data into a more structured and useful format.

• smell
• emotion
• sound
• taste
• color
• temperature
• texture
• tangible
• intangible
• dimension: ex. shape, geometry, spacing or layout

Prompt Expansion

A descriptor can take the form of abstract concepts or a discrete characteristic. The descriptor is typically manually interpreted from an Image or other descriptor. In some cases it may be beneficial to "expand" the descriptor into a set of more abstract descriptors, and then use those expanded descriptors in the pipeline, of course labelled appropriately.