by Damian Grimling | May 12, 2020 | Sentimenti research
The Sentimenti tools are evolving. At the moment we are entering into further and more advanced stages of tools development. In this article, however, we return to the roots of our project. It is worth mentioning what we actually measure and how do we understand particular emotions. And this is only the tip of an iceberg!
The easiest way to understand our concept is to use a ready-made description, created for the purposes of a certain project (whose Sentimenti tools are a tangible effect), developed by our researchers, Monika Riegel, PhD and Małgorzata Wierzba, PhD from Laboratory of Brain Imaging – Neurobiology Center, Polish Academy of Sciences.
Sentimenti tools. Definition of affective dimensions:
Valence, a sign of emotion and sentiment
- determines whether a given information or event evokes negative or positive emotions in us;
- has a range from negative emotions (caused by averse events) to positive emotions (caused by attractive events);
- the more positively we evaluate the information and events we experience, the more positive emotions are evoked in us;
- the more negative the information and events we experience, the more negative emotions evoke in us.
Emotional arousal
- determines the level of intensity of our emotions in relation to a given information or event;
- It ranges from no excitement (indifference) to strong excitement (agitation or excitement);
- Strong Arousal means a state of increased vigilance, attention and information processing;
- Arousal plays a key role in motivating our body to take certain actions;
- Arousal is also associated with specific physiological and neural responses (e.g. increased heart rate, accelerated breathing).
Basic emotions and their definitions
JOY
- Joy includes many positive emotions felt in response to what is known or new;
- we signal joy through a sincere, authentic smile, which consists of lifting the corners of the mouth diagonally upwards and tension in the circular muscles of the eye (lifting the cheeks and creating wrinkles around the eyes with age);
- joy is also shown by voice signals – breathing with relief or laughing or giggling;
- the main message of joy is “I like it”, so our support or encouragement for something.
SADNESS
- The feeling of sadness is a way to deal with loss and show others that we need support;
- Sadness signals include wrinkled lips (lower lip slightly raised and corners of the mouth lowered), inner corners of eyebrows joined and raised to the center of the forehead, raised cheeks;
- other manifestations of sadness are tears, as well as vocalisation expressing this emotion (weeping, trembling voice);
- The main message of sorrow is “comfort me”, and therefore an invitation to others to show us their support and care.
TRUST
- means believing that someone or something will behave in accordance with our expectations;
- feeling of trust brings us a sense of security and builds affection;
- The main message of trust is “I believe you won’t let me down”, it allows us to build not only intimate relationships with others, but also to find our place in society;
- Interestingly, we are more confident in faces similar to ours.
DISGUST
- The feeling of disgust with something allows us to avoid things that are harmful to us, both in the literal physical and mental sense;
- there are three elements of facial expressions expressing repulsion: the first is the ejection of the tongue reminiscent of spitting something out, the second is the lifting of the upper lip so that the gums and teeth are exposed, the third is the wrinkle of the nose and the expansion of the nostrils;
- the main message of disgust is to “go away from it”, which also signals to others to avoid the object of disgust because it is unhealthy, contaminated or reprehensible (socially or morally).
ANGER
- we feel anger when something blocks us or when we feel treated unfairly;
- when anger is uncontrollable, we raise our voice and scream, and when we have control over it, we take a sharp, attacking tone;
- the signal of anger on our face is a flash in our eyes, lowered eyebrows and squeezed lips;
- When people receive a signal of anger, they usually feel hurt and may try to take revenge by also showing anger;
- The main message of anger is to “get out of my way”, with a range from discontent to threat or attack, depending on the severity.
FEAR
- the fear of danger allows us to prepare for something that threatens us;
- and the most common signal of fear are wide open eyes, stretched lips and raised, joined eyebrows;
- the feeling of fear can also be accompanied by a reaction of avoidance. In example it is moving away from the source of fear, or dying;
- strong fear may be accompanied by an outburst of screaming, as well as signals such as heavy breathing, a slightly back-facing head and tense neck muscles;
- the main message of fear is “help!”, ranging from anxiety to panic, depending on the severity.
EXPECTATION
- an emotion involving excitement or anxiety in anticipation of upcoming events;
- Expectation is used to reduce the tension or stress associated with the challenge ahead by imagining it and developing a strategy to deal with it;
- The main message of the expectation is “I’m waiting for what will happen”. The ability to anticipate the effects of our actions in the future is essential for enjoying life.
SURPRISE
- emotion felt in reaction to unexpected events, expressing the discrepancy between our expectations and reality;
- the signs of surprise are raised and curved eyebrows, transverse wrinkles on the forehead, wide open eyes and enlarged pupils;
- is also visible through the lowered jaw, the separation of the upper and lower lips and teeth, the relaxation of the mouth
the surprise can be negative or positive;
- the main message of a surprise is “I didn’t expect it”. Although it ranges from light to very strong (the “run away or fight” reaction), depending on the intensity.
Emotion diads
Between 1960 and 1980, an American psychologist developed his theory of emotions. He decided to start with the eight basic emotions. According to Robert Plutchik’s theory of emotions, because it is referred to when different emotions are felt at the same time, they can create more complex types of emotions called diads. Diads arise from related but not opposing (mutually exclusive) emotions. The Sentimenti tools can analyze emotions based on 8 basic emotions. We stand out:
The basic diads (often felt):
- joy + trust → love
- trust + fear → humility, submissiveness
- fear + surprise → agitation, fear, horror
- surprise + sadness → disappointment
- sadness + disgust→ Repentance, repentance
- disgust + anger → contempt, envy
- anger + expectation → aggression, aggressiveness, aggressiveness
- expectation + joy → optimism
Secondary diads (sometimes felt):
- joy + fear → guilt
- trust + surprise → curiosity
- fear + sadness → despair
- surprise + revulsion → shock
- sadness + anger → suffering
- disgust + expectation → cynicism
- anger + joy → pride
- expectation + trust → fatality
Tertiary diads (less common):
- joy + surprise → admiration
- trust + sadness → sentimentalism
- fear + disgust → shame
- surprise + anger → indignation
- sadness + expectation → pessimism
- disgust + joy → pathology
- anger + trust → domination
- expectation + fear → anxiety
Opposites:
- joy + sadness → conflict
trust + disgust → conflict,
fear + anger → conflict
surprise + expectation → conflict
by Agnieszka Czoska | Apr 9, 2020 | Sentimenti research
Dr Jan Kocoń is a natural language engineer and the person behind the machine learning process within SentiTool, our solution for analyzing emotions in the text. Dr Kocoń coordinates the work of the linguistics team, integrates individual elements of the tool, and works closely with the IT team.
If you have to describe Sentimenti and the tools to anybody, what would you say first?
Sentimenti is a project meant to analyze emotions hidden in the text. Unlike competitive solutions that recognize the overtones of the text only (positive, neutral or negative), our tools manage to understand the text, assign specific meanings to the words in the text and name the certain emotions people feel about them. These emotions, in turn, provide the knowledge base for a machine learning mechanism that automatically recognizes emotions at the level of sentences and the whole text.
What does it mean that we analyse emotions in the text?
In the research carried out in our project we adapted the Plutchik model. It includes eight basic emotions: joy, sadness, trust, repulsion, expectation, fear, surprise and anger. We are able to estimate to what extent these emotions are expressed in the text.
How do we know what emotions people feel?
The knowledge base that helps our project includes more than 30.000 meanings of words, for which 20.000 unique respondents assign ratings for overtones and emotions. We are talking about “meanings” and not “words” on purpose, because words are ambiguous; for example “dark” means something different in “dark blue” or “dark people” and only in the latter case it carries emotions. Each meaning will ultimately receive 50 marks from different people. This allows us to know what feelings are evoked by certain meanings in the text. However, the emotion of the text is not a simple summation of the emotions assigned to the meanings in the text...
What else makes the emotion analysis tools in the text work?
Two things come to us to help. The first one is our gargantuan database of opinions. It came with associated overtones, derived from different areas: travel, medicine, products, services and more. We have over 10 millions of such texts in our database, which is an excellent source of information about the general feeling of the author. However, in order to find out what emotions a given text evokes in the reader, we also conduct our own research, analogous to research on single meanings.
This time the subject of these studies is the texts. The respondents attribute basic emotions to them, exactly the same way as they do with meanings of the words.
The second pillar of our Sentimenti tool is a combination of various machine learning methods. Experts in natural language processing provide us with tools for text analysis at the syntactic and semantic level, additionally they create rules for the analysis of meanings in context such as: negation, conjecture, weakening or strengthening of the overtones, etc. This is an additional help for automatic methods, such as deep neural networks, which are used to make the right conclusions about the emotions in the measured text.
What do you think automatic emotion analysis can be useful for?
Ultimately, I see many applications for our tools. The very first area that comes to my mind would be the marketing, or, more precisely, display advertising. This certain area covers the market of advertisements displayed in the context of web articles and is matching them with the emotions that the text of the publication evokes in readers. For example, in a sad text there could be an advertisement of an insurance company, and in a merry, joyful text there could be an advertisement for a trip.
Another area that we could cover is brand monitoring, i.e. analyzing how companies’ customers write on the Internet about a given company, its products and what emotions accompany them. Another interesting area could be sorting customers’ email complaints against the emotions contained in them, detecting conflicts arising in employee correspondence, detecting upcoming crises in Social Media, and even the possibility of diagnosing mental illnesses – the potential of Sentimenti tools is really huge!
What else do you plan to do in Sentimenti?
So far, there is a prototype ready with a simple text analysis on the level of meanings with an overtone analysis using our huge opinion resources. Currently in the Sentimenti team in Wroclaw I am managing to build a machine learning mechanism. It will make it possible to aggregate both information from the meaning knowledge base and information from the natural language processing stream. We are constantly receiving new data about the feelings of people reading certain texts, which are our teaching collection. The more data we gather, the better the quality of the tool there is.
by Agnieszka Czoska | Apr 9, 2020 | Sentimenti research
You’ve been in Sentimenti from the beginning. What was it like in 2016?
The business idea for the study of emotions in the text came from W3A.PL company from Poznan. After consultations with the environment of Poznań psychologists, cognitive scientists and linguists, a draft of the project for NCBiR (National Centre for Research and Development) was prepared and the search for subcontractors started. After estimating the market, it turned out that two units are capable of undertaking such advanced research work: LOBI IBD PAS and Language Technology Group of Wrocław University of Technology.
Once you got the grant, how did you start working?
As a research manager I was responsible for organizing the work of the team. It was important for me to combine the scientific teams of subcontractors and the business team into one team. The interface between business and science is not easy. In the Sentimenti team everyone – presidents, PhDs and MSc – speaks to each other by name, each person has the right to express their opinion and make decisions.
You are the research manager and scrum master of our team – how much did you have to learn to become one?
I learned the Scrum management methodology for R&D projects in the UK, where I worked in the Argument Analytics project conducted in cooperation with the University of Dundee and financed by Innovate UK, the British equivalent of NCBiR. I understood then that the key issue in the cooperation between science and business is good communication. A common team, preferably working in one place, frequent meetings and evaluation of results to check if this is really what we want – this is the heart of good projects. Many other R&D projects that I have observed did not achieve their goals precisely because of such a lack of communication.
How does the scrum method differ from your previous project experience?
I am a scientist and I have gained most of my experience in academic work and basic research. The transition to applied research was not easy, but I was given a lot by the British culture of openness, communication and respect – the values that are inscribed in Scrum and that we transfer to our team. The three pillars of Scrum are also important: transparency, inspection and adaptation. Transparency means that every person in the team – even new and unfamiliar with the subject – has access to all information (except, of course, confidential information). This helps a lot in overcoming crises, looking for a solution.
And what are inspection and adaptation?
An inspection is a frequent and short “review” meeting, during which we check what has already been completed, whether we do not have any obstacles that the project management should deal with, whether someone has too much or too little work. This helps to master the natural feature of research projects – unpredictability. When the results are different from we expected or when we get information from the business that a solution is not working – we can quickly adapt.
How do you see further development of Sentimenti?
In February, we have already finished our research work and moved on to development work, i.e. we use the collected knowledge and data in the work on Sentitol – our main tool for text analysis. Thanks to the fact that we use an iterative approach, we implement functionalities by adding them in subsequent versions of the product, and simultaneously – according to the Scrum methodology – we finish each Sprint (stage of work in Scrum) with a working product. At the moment, we have working software that recognizes eight emotions in texts in Polish, thanks to research on over 20 thousand people. This is already a solution that exceeds the scope of other solutions present on the market, and we are preparing two more versions.
In the next version of Sentimenti we will include a module using LSS (Lexical Syntactic Structures), i.e. elements of the language that affect the evaluation, e.g. good + no, + very, + a little. Then we will include a module that uses deep neural networks technology, or more precisely – BiLSTM (bidirectional long short-term memory neural networks), so that it can evaluate the emotions throughout the text immediately – and this is a unique solution on a Polish scale, but also worldwide. Our scientific publication about this module will be published soon.
Therefore, in the project we use fast prototyping, and in parallel to the work of the scientific team, the company implements any new solution for customers – because we have a great interest in our solutions. Thanks to this we have already achieved much better results (and faster) than we planned at the beginning.
by Agnieszka Czoska | Nov 8, 2019 | Conferences, Sentimenti research
Sentimenti = Emotions. We’ve previously discussed how to accurately analyze emotions with automated tools. Today, we’ll explain how we gathered the data that led to the creation of Sentimenti tools. This article is a guide to conducting effective research on the emotional meaning of texts.
The text was prepared for the GHOST Day machine learning conference and you can view the presentation in Polish.
Sentimenti. What emotions?
To train machine learning algorithms to automatically identify emotions in text, we first had to ask people how they feel. This seemingly simple question had to be broken down into several components.
First, what types of emotions should we consider? How many are there, and how do they differ? To answer these questions, we consulted the emotion specialists from the LOBI team. In psychology, there are various models of emotions, from simple to complex and multidimensional. We ultimately chose two models, which we now refer to as the sentiment and emotion models.
The sentiment model, based on Russell and Mehrabian’s 1977 paper, describes emotions along two axes: positive-negative and high-low arousal. As for the emotion model, we adopted the Plutchik model, both for its scientific robustness and because a portion of the Polish Slavic network had already been classified using it. This alignment allowed us to compare our findings with expert annotations, serving as a key test for accuracy.
What words?
Once we knew how to classify emotions, the next question was: what words to analyze? Our first step was focusing on the emotional meaning of words. We compared our findings with databases like WordNet and NAWL.
Our goal was to create a list of 30,000 words or meanings. Some words are ambiguous, with emotional tones shifting depending on context. For instance, “depression” can refer to both terrain and a mood disorder. We limited ourselves to a maximum of three meanings per word, each presented in context.
Thanks to the WordNet project, we learned that 27% of words have emotional meanings. These emotionally charged words took precedence in our analysis.
Sentimenti project: who participated?
To analyze emotional undertones in texts, we needed insights from a representative group of speakers. We worked with the nationwide research panel Ariadna to gather participants. Over 20,000 people took part in the study, providing data on at least 50 words each.
How We Collected Emotional Data
We designed a tool to assess word meanings on scales reflecting both sentiment and basic emotions. Participants evaluated words based on the emotional overtones they perceived in given phrases.
The study’s structure also considered participant fatigue. To ensure high-quality data, each person reviewed 150 words over three rounds, with breaks between rounds to maintain focus.
Beyond Words
Our next phase expanded beyond words to assess the emotional undertones of entire texts. Linguists have long known that the emotional meaning of a text goes beyond the sum of its words. The grammar and structure of a text also convey emotions.
For this phase, we analyzed existing reviews (e.g., hotels, doctors), as well as shorter forms like sentences and phrases. To ensure comparability with our word-level analysis, participants rated texts using the same emotional scales.
From People to AI
Sentimenti’s text classification tools now achieve high accuracy in identifying emotions, thanks to the solid dataset we built from word and text evaluations. While advanced neural networks may seem impressive, no AI can succeed without robust data to train on.
We’ve shared the details of our algorithm development both on our blog and in this scientific publication. Additionally, 20% of our word database will soon be published for researchers worldwide who study Polish emotions. This interactive database will have its own dedicated page, similar to NAWL’s list of affective words.
by Agnieszka Czoska | Jan 23, 2019 | Sentimenti research
In this read we want you to meet the idea behind the unique solution that we came up with. It works wider than popular sentiment analysis tools used today for monitoring internet activities and more. Why? Instead of measuring sentiment, we focus on emotions. What does the emotion analysis offer that sentiment analysis does not? Why do we need as many as 8 emotions to describe, or even predict somebody’s state of mind, moods, future actions etc.? And what could you possibly learn whilst knowing them? Let us find out.
What emotions does Sentimenti analyse?
In Sentimenti we have implemented our tools for using Robert Plutchik’s model of emotions. It reveals 8 atavistic, adaptive emotions that help a human being (and any other life form) to survive and create the base for developing far more complex emotional experiences.
The Plutchik model contains two positive emotions: joy and trust, two ambivalent ones: surprise and expectation, and 4 rather, so to speak, negative: sadness, fear, repulsion and anger. Each of them has its meaning, leading the person into an action, praise, criticism, need or surprise. We discuss these emotions one by one.
What is the result? By analysing all the 8 basic emotions, we can precisely assess how much your brand differs in the eye of Internet users, trace down customer opinions on products and services, find out on searched phrases, respond quickly to a communication crisis or compare the company to competitors, amongst the others. But sometimes we also love to have some fun checking on the emotional change during and after popular TV shows display, examining stock market reports reactions, an impact certain political, social or cultural event has and so on.
Sentiment analysis of joy and sadness
Joy and sadness are like black and white, sweet and sour or light and the darkness – a simple division into positive and negative.
Joy reflects a wide range of positive reactions and is related with a smile, the moment when we simply like something. Understood this way, it seems to be similar to a “positive sentiment” (as interpreted by sentiment analysis tools), but the key here is the intensity. Commonly understood joy (“positive sentiment”) is associated with a strong positive reaction to something. Our solution focuses on and lets you find out about any praise components that influence the levels of joy. Including the everyday courtesy.
Sadness is opposite. It describes the loss, a sense of loneliness and the need of comfort, sometimes is even being referred to the person’s ask for help. Together with the joy it forms a contrasting pair in Plutchik’s emotion circle.
The Plutchik model has the psychological background. According to the researches, expressing both emotions at the same time with a similar intensity proves an internal struggle, a very mixed attitude of the text’s author to the subject. It means the person is in unhealthy or somewhat uncomfortable, confusing circumstances. The comparable intensity of positive and negative sentiment cannot be interpreted this way.
The interpretation of joy and sadness levels in this case are different than the levels measured by sentiment analysis tools. In Sentimenti we use two models of text overtones, so we can measure it all. We also count the relationship between different emotions and sentiment. A correlation (we use the r-Pearson measure here) greater than 0.6 is called strong, and close to 0.9 means that the two variables are basically the same. We always measure the relationship between joy and positive sentiment and sadness and negative sentiment for broader cntext and more precise data collection. Depending on the subject, positive or negative sentiment can take almost entirely from a single emotion, but also as the result of many different emotions being expressed by Internet users at once.
The results concerning sadness and joy do not have to be directly related to those obtained in the analysis of sentiment – which also shows that it is the result of many different emotions being expressed in the texts at the same time. We are able to tell you what possibly is behind the joy and sadness.
6 other basic emotions
So now you know the joy and sadness are not just a simple sentiment, but can be developed from other, even complex emotions. We have taken a binary model and put I against the one with 8 categories. The two are already described, where are the other six? We will show what they do and why they differ from simple positive or negative sentiment.
Trust is expressing a sense of security, expecting something to happen the way we want to. It also covers a sense of community, a closer relationship, finding one’s place in the group, a feeling of mutual similarity etc.
Repulsion is an expression of disgust, strong criticism and aversion. When showing it, to us something is outrageous, alien or immoral. It is a strong emotion connected with the reaction of rejection and avoiding something. Many offensive nicknames that describe socially unacceptable groups are associated with repulsion.
Anger – another strong emotion, and very negative one. Instead of avoidance, it is associated with attack and sometimes leads into the aggression. In this way we express threats, but also dissatisfaction. It is accompanied by curses and shouting (in speech, in writing it can be expressed with stronger words).
Fear, as in life, is accompanied by danger, threat and risk. It is associated with escape, a desire to hide. It is also a cry for help.
Surprise comes up when our predictions don’t come true or something unexpected just happens to us. When combined with joy, it gives admiration, and with sadness – disappointment. It can have very different intensity, from a slight surprise to (also associated with fear) panic.
Expectation – the last of the emotions we describe. The opposite of surprise because it means making predictions and believing that they will come true. It can therefore be associated with anxiety or excitement, so it is also ambivalent.
8 emotions make sense
Describing the overtones of a text, portal or brand awareness on the scales of 8 emotions gives you a lot of information. Sometimes only three emotions from the whole model may matter the most, but you will only learn this when having an insight into the whole experience’s spectrum.
Emotions in the titles of portals are very different in terms of the sound of the start page (more in this entry)
First, the sentiment analysis is something different from the emotion analysis, yet we are talking here about two complementing models. Based on the results of one of them, one can predict to some extent the results of the other. Emotional analysis offers a much better insight into what actually checked texts communicate and what kind of consequences they may bring on. For example, an anger is a better catalyst for conflict or popularity loss than sadness. A decrease in trust is more dangerous than a decrease in joy. The graph above shows how different emotions are placed on information portals depending on their character and target group.
Individual emotions are not only differently described by us, but also differ qualitatively in the reader’s perception of the text (we examined it on 22 thousand people) and in the intention of its author. More trust does not necessarily mean more joy – these are very different feelings and can sometimes turn out to be mutually exclusive. These are basic emotions, so they should be independent of each other. At the same time, they form a system and sometimes interact, even if one displaces the other. If you want to know more about emotion analysis, we have also written about go od practices to draw valuable knowledge from measuring emotions with Sentimenti tools.
