“Sentino is an AI-powered semantic text analysis tool built to provide comprehensive psychological portrait of target audience. It is served as an API.”Source: sentino.org
But what exactly is meant by this wording? Our development, marketing and sales specialists are frequently asked the following questions related to our services and their underlying technologies:
- What is the psychological basis of Sentino analysis?
- How do you know that psychological data provided by Sentino are valid?
- What IT technologies is Sentino analysis based on? How does Sentino analyze the personality of individuals?
- What is Statistical Confidence Level?
- What is (after all) an API?
This article is purposefully written to give brief answers to all these topical questions.
1. What is the psychological basis of Sentino analysis?
Psychological basis of the Sentino solution is represented by the selected psychological personality tests like BIG 5, NEO, RIASEC, ORVIS, ORAIS, DISC, etc. These inventories are carefully chosen according to several basic criteria:
- Scientific validity, relevance, and reliability proven by a large number of studies duly performed according to standard requirements and methodological principles and published in reputable scientific psychological journals.
- Availability of vast open-source data constituted by answers of respondents from various groups (that differ by their age, gender, nationality, etc.) to the questions of inventories. Depending on the questionnaire, from 1,000 to 1,000,000 individuals have provided their answers. More than 3,000 questions were answered. Most of the respondents are people from WEIRD (western, educated, industrialized, rich, and democratic) countries. However, this is a common limitation inherent for almost any psychological test and reported by many scientists.
- Popularity of questionnaires and their widespread use by psychologists, HR specialists, academic community, consultants, and people interested in self-understanding.
2. How do you know that psychological data provided by Sentino are valid?
The psychological data underlying Sentino AI models are valid since we use only scientifically proven and reliable psychological scales. In terms of psychometrics, we use generally accepted standard procedures that are not able to somehow introduce any invalidity.
The unique and innovative character of Sentino solution is determined by our proprietary AI-based approach to semantic text analysis. Sentino Personality AI ensures the extraction of psychologically significant information from any test at the high level of accuracy. It is also capable of detection of intents and sentiments. Some minor discrepancies may occur in the analysis course due to the following fact. Artificial Intelligence always searches its databases looking for the psychologically significant fragment that most closely matches the text fragment under analysis. It may be difficult to find if the analyzed text fragment is provided by a person with a low level of language proficiency or somebody who prefers long/complex/ambiguous phrases. However, this problem can be solved via continuous research and training of models. And we are constantly working on its resolution.
3. What IT technologies is Sentino analysis based on? How does Sentino analyze the personality of individuals?
The Sentino solution comprises AI / ML / NLP technologies. Our development team has integrated vast psychological data and built a natural language interface for it. We have trained our models using the state-of-art NLP stack (e.g. Spacy, Transformer) that is able to transform any structured or unstructured natural text (self-descriptions, CVs, answers to questions, LinkedIn profiles, social media posts, messages, etc.) into psychologically meaningful representation and link it to validated inventories (please see Question No. 1 above). We continuously keep researching and updating the models, striving for higher accuracy and better match with the customers’ needs.
Sentino Personality AI uses the available psychological Big Data (the arrays of the respondents’ answers) to create a multidimensional psychological vector space. It consists of sentences and segments containing psychological information and grouped by personality traits and facets (according to the structure of the correspondent questionnaires).
Any text related to the individual under analysis undergoes the following processing steps:
- Division of the whole text into sentences
- Further subdivision of sentences into segments
- Segment classification (e.g. distinguishing of demographic and psychological data, as well as other significant topics)
- Generation of the individual’s personality portrait based on the psychological information derived
At that, the individual’s personality is represented as a vector within the psychological vector space and can be easily projected onto each personality trait or facet of interest.
4. What is Statistical Confidence Level?
Here, in Sentino, we do our best to exclude the “black-box principle” when developing our AI/ML models. We strive for transparency required for trustworthy and explainable predictive analysis and for this reason introduce the so-called “statistical confidence level”. This statistical coefficient of reliability is always calculated for each numerical value or result provided by Sentino Personality AI. It shows the reliability of the correspondent result within the range from 0 to 1. The coefficient may decrease if your data is insufficient or inconsistent (e.g. you have mutually exclusive answers).
5. What is (after all) an API?
An application programming interface (API) is a way for two or more computer programs to communicate with each other. It is a type of software interface, which facilitates the use of functions of one application within another. In contrast to a user interface, which connects a computer to a person, an API connects pieces of software to each other. It is not intended to be used directly by a person (the end user) other than a computer programmer who is incorporating it into the software. APIs help developers to expand the functionality of their products and connect them with others.
By the default, Sentino is served as an API that users our own IT facilities. By the request, we are ready to deploy our comprehensive turnkey solution directly at the premises of your company or organisation (our “Run on Premise” service package). We install API clients on your computers and perform all the necessary works related to API configuration, setting, and testing in real working environment.
Moreover, we have developed dedicated technical solutions for our top-selling products – Interview Semantic Analysis API and LinkedIn Personality API. These solutions comprise a user-friendly interface, Sentino Dashboard, that is operated just in several clicks.