Everything You Need to Know about Online Text Analysis

Analyzing and understanding all the text your company receives each day can be challenging, especially when most of that data comes in the form of unstructured text. Be it tweets, online reviews, emails, or surveys, all this information soon starts to pile up. Even if you have a superb, experienced team to sort this data, it would take them hours and even days to read and organize all the messages your company receives through various online channels. On top of that, it’s an unrewarding and monotonous task. 

Luckily, text analysis with online tools can help your team sort, categorize, and gain deep insights from your data (customer reviews, emails, chats, etc) in a quick and efficient way. And don’t worry, this AI tool will never get tired or bored. Not only are text analysis models easy to train and use, but they will also save your company considerable resources.

In this post, we will describe how text analysis with machine learning can be useful for your company. We’ll show you how easy it is to get started by walking you through some online demos for text analysis, then we’ll explain how machine learning can optimize your resources. Finally, we’ll introduce you to some of the best online text analysis tools, and even show you how to build a custom model to analyze your text. 

Are you ready? Then let’s go!

What is Text Analysis?

Text analysis is an automated process that uses machine learning to extract or classify information from texts, such as Facebook posts, chat messages, Google Store reviews, surveys, and more. It performs faster than human agents, allowing you to structure your data in seconds and obtain interesting insights about your company. 

For text analysis tools to be able to sort texts, however, you need to train them. In other words, you need to teach machine learning algorithms how to carry out certain tasks. Some of them involve analyzing and understanding the underlying meaning of a document, while others pluck out information from existing text. It may sound complicated, but it isn’t if you are using online text analysis tools like MonkeyLearn.

Text analysis models learn in a very similar way to human beings. To be able to distinguish a certain object, let’s say a car, a child needs to see it first and recognize it’s primary features, such as the wheels, doors, seats, etc. After understanding all the parts that make up a car, a child will be able to see a car in the street and differentiate it from a bike, even if this car isn’t the same one the child had originally seen.

Machine learning models work in a similar way. When using MonkeyLearn’s models, the first thing you’ll need to do is upload a file with the texts (for example, product reviews) you want to analyze. Then, you’ll need to tag those reviews manually. The more samples you tag, the smarter your model becomes. When it reaches a certain number of tagged texts, the algorithm will start differentiating texts and making predictions on its own.

MonkeyLearn has several pre-trained models ready for you to try. These include sentiment analysis, topic analysis, keyword and entity extraction, urgency detection, language detection, and more. These models serve different purposes, and you should choose the one that’s more appropriate for the problem you are trying to solve. You’ll find more information about each of these models later on. Now, let’s discuss the importance of text analysis for your company. 

Why is Text Analysis Important?

You might be wondering how text analysis can help your business. Well, there are many areas in which you can implement online text analysis tools, from customer support and marketing to product and dev teams.

For example, your teams might need to extract specific information that is already in text data, like company names, keywords, or pricing. Or you might want to figure out the sentiment of a batch of customer survey responses and classify each one as Positive, Negative, or Neutral. Both of these tasks can be completed in a matter of minutes using online tools.

Now, you might be thinking, why not just ask your team to manually analyze the data you receive. One, this process is repetitive and time-consuming, and two, it’s inconvenient for your company. Take Paypal as an example, which has more than 270 million active accounts! Each time customers’ problems are solved, Paypal asks them to fill in a satisfaction survey. This generates tonnes of raw data to be analyzed every time they close a ticket!

It would take ages for a person to analyze these responses manually. Also, as soon as teams have finished examining one batch of responses, hundreds of new ones are likely to have appeared. With text analysis, you can automate processes like this and analyze vast amounts of data in real-time, faster, and more accurately than using human agents. Let’s take a look at some of the benefits in more detail:

  • Scalability: It’s not the same manually analyzing 10 reviews as it is 100. It will take humans a lot more time and effort, and you may lack the resources if there’s a sudden influx in product reviews. Online text analysis tools, in turn, are scalable and can analyze vast amounts of information in just seconds. Imagine that you suddenly start getting a lot of complaints, and you don’t have the manpower to handle them. Instead of hiring new employees, you can use online text analysis tools such as MonkeyLearn to give you a hand.

  • Real-time analysis: 64% of consumers expect real-time communication with companies. This is quite difficult to achieve, especially if your volume of customer support tickets is high. Text analysis, however, can sort issues and detect urgent comments that need an immediate answer in real-time, 24/7. That means your teams won’t have to spend hours sorting tickets and can take immediate action on urgent issues – potentially avoiding a crisis.

  • Consistent Criteria: It’s in our nature to make mistakes. Even more so when tasks are not motivating. Machines, on the other hand, never get tired or bored. The results from analyzing texts with AI will be more accurate and thorough than results obtained from analyzing texts with human agents. In addition, online tools like MonkeyLearn, are not influenced by personal experiences or beliefs when tagging texts. They use consistent criteria to analyze data, no matter the volume, rendering the results reliable and accurate.

Demos and Techniques for Text Analysis

MonkeyLearn has a number of pre-trained models ready for you to try out. Let’s take a look at some of these online demos below, and examine how they can suit your needs!

Text Classification

Text classification refers to the process of sorting text through automated tagging. In other words, machine learning organizes texts into categories so that it’s easier to make sense of them. Some of the most common models for text classification include sentiment analysis, intent detection, topic modeling, and language detection, among others. 

Sentiment Analysis

This model tries to detect the emotions people express with their words, and classifies texts into Positive, Negative, or Neutral. One of your reviews might read “Great timing when answering queries!”. If you are using a sentiment analysis model, then you would tag this review as Positive. If you have a comment such as “It’s too expensive”, you would tag it as Negative

By using a sentiment analysis tool, you can easily assess your brand’s reputation in media outlets, social media, and more. This machine learning model can also improve your products, by examining customer feedback or identifying urgent issues.

Try out our pre-trained model for sentiment analysis. Just type something positive or negative, and see how it goes!

Topic Analysis

Topic analysis is another example of text classification, and it helps you sort texts by topic. For example, you can classify product reviews using tags such as Pricing, Customer Support, Design, and more, and automatically route them to team members best equipped to deal with each topic. 

You can even combine this model with a sentiment analysis model to find out how customers feel about a specific topic. You can try this pre-trained model for classifying NPS responses for SaaS products by typing your own text to see how it’s categorized. 

Language Detection

Text analysis can also detect the language of a text. This is very useful when you have issues from customers in multiple languages and you want to route incoming messages to agents who are able to understand them. Our pre-trained language detector can differentiate between 49 different languages. Give it a try!

Intent Detection

Analyzing the reason why a person sent a message is crucial to truly understand customer interactions. This type of classifier can help you see if the client has the intention of buying your product or if he wants to complain about it. Let’s say you receive an email. An intent detection model will analyze the body and the subject of the email, and depending on the words and expressions customers use, machine learning will tag the message accordingly. A client who writes “Let’s arrange a call tomorrow” is clearly Interested, and your text analysis tool will recognize that intent. 

Check out this pre-trained intent detection model. Write or paste emails from customers into the text box, and find out if a customer is interested, not interested, if they want to unsubscribe, and more.

Text Extraction

This is another form of automated language processing. This text analysis technique will extract specific information that is already in the text. For example, you can examine a batch of product reviews and obtain the main keywords customers use when writing them. 

Afterwards, you may use those key phrases and terms to organize them, create graphs, and solve issues. You can even create your own custom model to extract specific pieces of data you are interested in. You can find more details about text extraction models below. 

Keyword Extraction

If you are looking for the most relevant terms or phrases, then a keyword extractor can help you. When using this model, you’ll get a sort of summary of your texts in list form. Then, you can use this information to index data, or to create a custom classification model using these keywords as tags.

Take a look at how our pre-trained keyword extractor works. 

Entity Extraction

This model will analyze your text and extract entities that already appear in the text. With our pre-trained entity extractors, you’ll be able to pull out the names of people, brands, or places in seconds. Check out this pre-trained entity extractor and try it for yourself!

Text Analysis Online Tools

There are many text analysis tools available, including open-source libraries such as TensorFlow, NLTK, PyTorch, and Scikit-learn. You can build your own models using these tools, although you’ll need to know how to code to successfully build a text analysis model using this open-source software. 

There are also online tools for text analysis, like MonkeyLearn, which have clear advantages over open source ones. Let’s take a look at some of these advantages below.

  • No Setup: When using online tools, you won’t have to spend hours setting up the required ecosystems and libraries. For example, before you can start using PyTorch you’ll need to install Python and all its dependencies (such as Numpy, Setup Tools or PyYAML (this also involves installing the language it was built with – in other words, the language it understands. In this case, PyTorch was built using Python so you’ll need to be familiar with Python in order to use PyTorch. 
  • No Code: You don’t need to know how to code to get started with online tools. You just have to feed them texts and they’ll return a prediction automatically. So much easier than building something from scratch, right?

  • Easy Integration: If you want to integrate text analysis tools with other apps like Zapier or Google Sheets, that’s possible (and easy!). For example, you can connect MonkeyLearn with Google Sheets using its seamless integration – no coding or programming knowledge needed!

  • Pre-trained models: online tools usually offer pre-trained models that you can use right away. Using open-source tools, on the contrary, usually means you need to train your own model, which is time-consuming. MonkeyLearn offers an array of pre-trained models that you can use to get started with text analysis right away.

Here are some of the best online text analysis tools available:

  • MonkeyLearn
  • Google Cloud NLP
  • IBM Watson
  • Lexalytics
  • MeaningCloud
  • Amazon Comprehend
  • Aylien

If you are in doubt whether these tools truly suit your needs, you can create an account and easily test them for free.

Tutorial: How to Get Started with an Online Tool for Text Analysis

Let’s imagine you have just carried out a survey to find out how satisfied your customers are with a new product you launched last month. You get a bunch of answers, which is great! But there are many more than you were expecting. Don’t worry, you won’t need to spend endless hours reading your customers’ responses one by one.

Instead, you can try our pre-trained sentiment analysis model. You have several options – either type text in the box, upload a batch of responses you have in Excel, or even integrate it with other tools, such as Google Sheets. Let’s see the different ways in which you can use this model in MonkeyLearn!

a) Try Out a Model via MonkeyLearn’s Simple User Interface 

Access your dashboard and go to this sentiment analysis model. You’ll see a text box in which you can add a customer comment to see how it works:

b) Upload a File to Run Text Analysis on Large Quantities of Data

If you have an Excel file, for example, in which you have compiled many reviews, you can upload it to MonkeyLearn in seconds.

Once you have accessed the sentiment analysis model, click ‘Batch’ on the left-hand side of the screen. There, you can upload an Excel or CSV file containing your texts for the model to analyze. Just select the rows to examine and voilá! After a few seconds, you will be able to download an Excel file with the results (Positive, Negative, or Neutral) next to each review:

c) …Or, Integrate The Model With Your Apps

There are different integrations available so you can easily connect a model in MonkeyLearn with your apps. Just connect MonkeyLearn with the software you want to enhance with a text analysis models, and you’ll be able to analyze incoming messages in the apps you use, without coding or programming:

If you know how to code, you can use MonkeyLearn’s API to run text analysis models with programming languages such as Python, Ruby, PHP, or Java. 

Create a Custom Text Analysis Model 

If you are looking for more accuracy when examining your texts or use your own set of tags, then you could build and train a custom model for text analysis. This model will learn from your own data and criteria for extraction or classification. Let’s take a look at how to build a custom model for topic analysis. It’s super easy! 

1- Let’s Create a Text Classifier

Access your dashboard and click on create a model. In this example, we’re going to build a classifier:

2- Upload Your Texts

In this step, you’ll need to upload some texts as samples to train your model. We’re going to upload a batch of product reviews, for example. You can upload either an Excel or CSV file, or just import texts from apps like Zendesk, Twitter, Gmail, or RSS feeds. 

3- Define Your Tags

The next thing you need to do is define your tags. You’ll need at least two. Take into account that the more tags you add, the more samples you’ll need to train your model: 

4- Start Tagging Samples

Now it’s time to tag your reviews to train your model. The more you tag, the smarter your model will become, and the more accurate it will be at making predictions on its own:

5- Test the Model! 

Once you have finished tagging, you’ll see a screen like this one:

Click the ‘test’ button and see how it detects topics by typing your own review in the text box: 

If you want to improve the accuracy and confidence of your model, you just need to keep adding and tagging new data samples. You can also correct incorrectly tagged examples within your model. If you’d like to know a bit more about how online text analysis tools work, check out this guide on data analysis.

Use Cases & Applications

Text analysis is very useful when automating business processes and getting insights for better decision making. Let’s see some examples of how this automated process can help your company, especially when examining customer feedback and improving customer support.

Customer Feedback

Analyzing product feedback you obtained via NPS or satisfaction surveys can be a nightmare. Open answers can offer a lot of crucial information about your new product, for example, but this data is not easy to analyze. A person might spend hours just reading the answers, not to mention trying to understand the insights! 

Machine learning, on the other hand, is fast and reliable when trying to understand what your customers like and dislike about your brand. For example, you can do what Promoter.io did, and use this pre-trained online keyword extractor to understand the main topics mentioned in customer reviews. If you combine it with sentiment analysis, you’ll discover if customers are talking about these topics positively or negatively.

It’s also possible to analyze online reviews for unfiltered customer feedback. Picture this: you work for a company like Salesforce and would like to know what features your customers like and don’t like about the platform. The first thing you need to do is gather reviews from review sites such as Trustpilot or Capterra. Then, you can run a text analysis model (such as sentiment analysis) to see how people feel about the company. Finally, you can examine product reviews of your brand and compare those with the competition. 

Let’s take a look at this example, below, where we compared the sentiment of tweets for four different telcos: AT&T, Verizon, Sprint, and T-Mobile. After running sentiment analysis on each company’s Twitter mentions, we discovered the following: 

Percentage of positive tweets mentioning a telco.

T-Mobile is definitely winning, with nearly 20% of their mentions being classified as Positive. Sprint comes in second, with 15% of positive mentions. What’s interesting here is that the two companies that have the least positive tweets are the largest ones.

Customer Support 

Improving the customer experience is more important than ever. Fifty-four percent of customers have higher expectations of customer service today compared to last year. Machine learning, then, can be the helping hand you need. Of course, online text analysis tools won’t replace the human workforce, but they can help solve certain tasks that take a lot of time. For example, you can integrate text analysis tools like MonkeyLearn with customer service help desks and tag all your incoming tickets as soon as they arrive! 

Normally, a team member would have to categorize these queries manually, and that’s before solving the issue. Yes, it seems like an easy task, but it’s monotonous and time-consuming if you have hundreds or thousands of customer queries to go through. On top of that, the criteria that team members follow when tagging tickets might not always be consistent. In this case, delegating the ticket tagging and triage process to online text analysis tools is a smart choice. 

Using machine learning tools will also help you detect disgruntled customers and immediately route their issues to the top of your queue. Not all the messages you receive need an immediate solution. But, how do you detect the urgent ones without wasting valuable time reading each email or ticket? Well, text analysis can define how urgent (or not) each customer issue is, and tag it accordingly. 

In customer support, a lot of time is wasted bouncing tickets from one team to the next, especially if you’re an international organization and receive customer issues in various languages Using text analysis models, such as this pre-trained language classifier, you can make your routing process a lot more efficient. As soon as you get an email, the model will tag it with the corresponding language and send it to the agents best equipped to respond. 

You can also use analytics to gain deep insights into what’s happening across your customer support team. The most common way to assess how your team is performing is by measuring their response times, how long it takes them to solve issues, and how satisfied customers are. This last element is the hardest one to measure. Is there any efficient way to get insights from your customers’ messages? Well, sentiment analysis can give you a hand and examine your texts to understand the feeling behind support tickets, chats, emails, and surveys.

Final Words

Analyzing data on your own can be a drag. Tagging customer support tickets or survey responses is tedious and time-consuming. Especially if it’s something your team has to do every day and in high volumes. But don’t worry! 

Thanks to text analysis, your team won’t have to spend endless hours reading text to determine their urgency, topic, sentiment, and so on. Machine learning can take on this tedious task, and since text analysis is scalable, fast, and consistent, it will provide efficient support to your human workforce. 

The best part is that there are plenty of machine learning tools available online, and it’s free to get started. Plus, with tools like MonkeyLearn, you won’t need to enter a single line of code. 

We’ve mentioned several ways in which online text analysis tools can help your business make the most of your data. Now, it’s over to you. Give text analysis a try and ask for a demo today! Our team will be more than happy to show you how to use MonkeyLearn’s text analysis tools. 

Federico Pascual

Federico Pascual

COO & Co-Founder @MonkeyLearn. Machine Learning. @500startups B14. @Galvanize SoMa. TEDxDurazno Speaker. Wannabe musician and traveler.


Have something to say?

Text Analysis with Machine Learning

Turn tweets, emails, documents, webpages and more into actionable data. Automate
business processes and save hours of manual data processing.