What do you do before purchasing something that costs more than a pack of gum? Whether you want to treat yourself to new sneakers, a laptop, or an overseas tour, processing an order without checking out similar products or offers and reading reviews doesn’t make much sense anymore. Thanks to comment sections on eCommerce sites, social nets, review platforms, or dedicated forums, you can learn a ton about a product or service and evaluate whether it’s a good value for money. Other customers, including your potential clients, will do all the above.
People’s desire to engage with businesses and the overall brand perception depends heavily on public opinion. According to a survey by Podium, 93 percent of consumers say that online reviews influence their buying decisions. Users may not give you a chance once they’ve read a few bad reviews. They won’t research whether feedback was fake or not. They’ll choose another option. In this context, organizations that constantly monitor their reputation can timely address issues and improve operations based on feedback. Sentiment analysis allows for effectively measuring people’s attitude towards an organization in the information age.
Sentiment analysis is a type of text research aka mining. It applies a mix of statistics, natural language processing (NLP), and machine learning to identify and extract subjective information from text files, for instance, a reviewer’s feelings, thoughts, judgments, or assessments about a particular topic, event, or a company and its activities as mentioned above. This analysis type is also known as opinion mining (with a focus on extraction) or affective rating. Some specialists use the terms sentiment classification and extraction as well. Regardless of the name, the goal of sentiment analysis is the same: to know a user or audience opinion on a target object by analyzing a vast amount of text from various sources.
You can analyze text on different levels of detail, and the detail level depends on your goals. For example, you may define an average emotional tone of a group of reviews to know what percentage of customers liked your new clothing collection. If you need to know what visitors like or dislike about a specific garment and why, or whether they compare it with similar items by other brands, you’ll need to analyze each review sentence with a focus on specific aspects and use or specific keywords.
Depending on the scale, two analysis types can be used: coarse-grained and fine-grained. Coarse-grained analysis allows for defining a sentiment on a document or sentence level. And with fine-grained analysis, you can extract a sentiment in each of the sentence parts.
This analysis type is done on document and sentence levels. In fact, most specialists use it to analyze sentences rather than whole documents. Coarse-grained SA entails two coherent tasks: subjectivity classification and sentiment detection and classification.
1. Subjectivity classification. First, it’s necessary to determine whether a sentence is objective or subjective. An objective sentence contains some facts about an object or topic: Three strangers are reunited by astonishing coincidence after being born identical triplets, separated at birth, and adopted by three different families.
A subjective sentence, as the name suggests, expresses someone’s attitude regarding a subject: This apartment is wonderful. I enjoy every minute I spend in here.
2. Sentiment detection and classification. The goal of this operation is to define whether a sentence has a sentiment or not and if it does, to determine whether the emotion is positive, negative, or neutral.
https://www.altexsoft.com/media/2018/09/sentiment_analysis-300x225.jpg 300w" sizes="(max-width: 683px) 100vw, 683px" />
Sentiment by polarity. Source: KDNuggets
Sometimes people share their points of view without emotions. For instance, the author of the sentence I think everyone deserves a second chance expresses their subjective opinion. However, it’s hard to understand how exactly the writer feels about everyone. So, the sentence doesn’t express a sentiment and is neutral. Neutral sentences – the ones that lack sentiment – belong to a standalone category that should not be considered as something in-between.
Let’s look at this comment: One of the most surprising and satisfying movies of the year. According to the phrase, the reviewer enjoyed the movie, so this sentence contains a positive sentiment.
And the following review is a clear example of a subjective sentence with negative sentiment: The fact that it’s also clumsily made and rife with mediocre performances seems almost beside the point in the context of how pointless this thing is in the first place.
However, objective sentences can also express a sentiment: I bought this waterproof camera case because it’s meant to be more reliable than a standard one. It’s clear from the context that the case wasn’t what the person expected. The sentence has a negative sentiment, but it’s expressed implicitly.
Sentiment doesn’t depend on subjectivity or objectivity, which can complicate the analysis. But we still need to distinguish sentences with expressed emotions, evaluations, or attitudes from those that don’t contain them to gain valuable insights from feedback data.
The devil is in the details, as they say. If you need more precise results, you can use fine-grained analysis.
You apply fine-grained analysis on a sub-sentence level and it is meant to identify a target (topic) of a sentiment. A sentence is broken into phrases or clauses, and each part is analyzed in a connection with others. Simply put, you can identify who talks about a product and what exactly a person talks about in their feedback. In addition, it helps to understand why a writer evaluates it in a certain way.
The fine-grained analysis is useful, for example, for processing comparative expressions (e.g. Samsung is way better than iPhone) or short social media posts.
Not only does it allow you to understand how people evaluate your product or service, it also identifies which feature or aspect they discuss: A touchpad on my laptop stopped working after 4 months of use. This way, you know exactly what must be improved or reconsidered.
The capability to define sentiment intensity is another advantage of fine-grained analysis. In addition to three sentiment scores (negative, neutral, and positive), you can use very positive and very negative categories.
Sentiment analysis allows you to look at your operations from a customer point of view. But how do you extract that knowledge from user-generated data?
Data collection and preparation. First, you need to gather all relevant brand mentions in one document. Consider selection criteria – should these mentions be time-limited, use only one language, come from a specific location, etc. Then data must be prepared for analysis: one has to read it, delete all non-textual content, fix grammar mistakes or typos, exclude all irrelevant content like information about reviewers, etc. Once we have data prepared, we can analyze it and extract sentiment from it.
As dozens or even hundreds of thousands of mentions may require analysis, the best practice is to automate this tedious work with software.