Text mining: The business use primer

· Thomas Wood
Text mining: The business use primer

Your NLP Career Awaits!

Ready to take the next step in your NLP journey? Connect with top employers seeking talent in natural language processing. Discover your dream job!

Find Your Dream Job

Text Mining: Quick overview

The modern enterprise has access to vast amounts of unstructured data but that data can only prove useful if the desired insights can be extracted from it.

This is where text mining tools and text mining APIs come in, a way to make these vast quantities of unstructured data accessible, useful, and insight-worthy to drive business decisions.

When businesses are able to gain the right insight through all this unstructured data, they can generate much deeper insight into, for instance, what their customers are thinking or saying, and therefore, make very smart business decisions across not just customer service, for instance, but multiple use cases, as we will soon discover.

What is text mining?

Text mining has now become an essential capability of businesses of all scales and across nearly all sectors. It uses techniques like entity extraction or entity recognition, text classification, and sentiment analysis to acquire highly useful information and knowledge which may otherwise be hidden within unstructured textual content. If you were to find a needle in a haystack knowing that the ‘needle’ could really help drive your business to the next level, wouldn’t you do everything to find it? That’s precisely one of the functions of text mining.

In the business world, AI and text mining tools enable organisations to uncover specific patterns, trends, and insights from very large volumes of raw or unstructured data. This capability of pushing aside all non-relevant or unimportant data to acquire highly valuable answers has led to the rapid adoption of text mining tools and techniques throughout all organisational levels.

At the core of text mining are AI (artificial intelligence) techniques which help to automatically discover trends, patterns and other valuable information from unstructured text, as we already know. This text may come from survey responses, support tickets, articles, websites, customer feedback, emails, text messages, and other sources.

Since manually scanning and classifying each one of these documents to find your “needle in a haystack” would simply not be practical, cost-effective or time-efficient, text mining can save you a considerable amount of time and effort – the discoveries you make through a reliable text mining API can not only help you make better decisions but also drive action where action needs to be taken.

This should answer the “what is text mining” part. Now, let’s shift our ttention to more important aspects of the topic.

Why is text mining important and how can it be used in business?

Industry experts believe that as much as 80% of business data contains unstructured information, including text.

Text mining tools allow businesses to extract highly valuable information from unstructured text which, as you know, is generated in massive volumes in the form of customer service tickets, social media posts, email messages, chatbots, and what have you.

When there is an automated process in place, such as text mining, businesses no longer have to go through the ordeal of having to manually sift through every piece of textual (or other form) data to analyse information that could potentially serve in improving products and services, or help in making key business decisions, for example.

Apart from the time factor and how utterly frustrating it can be to analyse each text record manually, automatic processing of text documents can quickly produce far more accurate and consistent results. Therefore, no business should treat text mining as a ‘nice to have’ – on the contrary, it has become a must-have for any business that wishes to grow, progress, and remain competitive.

Furthermore, text mining can help businesses discover and respond to problems they may be experiencing in, say, customer service, production or manufacturing, anticipate threats to the business (either from competitors or otherwise) – or, provide their customers and stakeholders a more streamlined and personalised service, for example. The possibilities are nearly endless.

Business use cases of text mining + a few examples

To give you a better idea of how text mining APIs can help boost usiness, here are some use case examples:

Reputation management

Cancel culture has, unfortunately, made its way into the business world as well, although it’s nothing new. However, what this means for businesses operating in any given sector is that must do everything possible to ensure that they have a beaming public image.

Text mining lets you understand raw data captured from VoC (voice of customer) programs or social media listening, for example. It will analyse everything from news articles, comments, tweets, and other forms of feedback to detect whether anything negative has been mentioned about your business, or whether anyone posting and/or spreading that negative feedback is linked to it.

This may include everyone from company officials and employees hemselves to customers and partners, as well as the groups, political arties, and organisations your business supports. With the right text ining tools in place, you can easily take the appropriate steps in real ime to mitigate a potential reputational crisis before it has a chance o wreak havoc.

Search engine optimisation

Popular search engines like Google and Bing use text mining APIs to identify spam and filler content in websites that claim to be doing nothing more than ‘content marketing’, for example. This is just one example of how text mining can be used for search engine optimisation purposes, as it helps Google identify specific context, intent, and spelling variations to mark a website’s content as ‘spam’ – or to red flag a website that, for instance, has been using unethical/blackhat SEO techniques to artificially boost its SEO ranking.

Text mining can also be used to ‘fine tune’ an organisation’s internal search engine by integrating a text mining API which allows authorised personnel to semantically search every company document to dig out specific information. For instance, you could use text mining tools integrated into your video search to analyse text overlays, words, logos, and images across your entire corporate video library by identifying specific subject matters, topics, and themes. This can be very useful when sending a certain portfolio or product video to a potential customer or, for example, when you want to use something as evidence when defending your company in court. The use cases are wide and varied, either way.

Fast Data Science - London

Need a business solution?

NLP, ML and data science leader since 2016 - get in touch for an NLP consulting session.

Threat detection and cyberbullying

The integration of text mining tools can help to ensure that your community standards are complied with every time someone leaves a tweet or comment on your social channels. For example, specific keywords may be categorised as ‘threats’ while your text mining API will automatically pick them up, including misspelled words and synonyms people may be using to try and ‘cheat’ the algorithm. The same principle can be easily applied to all kinds of online business communications, such as emails.

This kind of integration can prove to be very useful in cases involving cyberbullying as well as those pertaining to public security measures – during public events, for example, to ensure everyone’s safety, security, and peace of mind. In fact, video and audio content may also be analysed at the same time to detect any threats or policy violations. In case of cyberbullying, specific kinds of texts or phrases can be analysed to ban people from using a social media service or even to track down their whereabouts in more serious cases.

Finding historical and current patterns in data has become very important in highly monetised and time-sensitive use cases, such as new product development, clinical trials, medical treatments and real estate planning.

Text mining can allow organisations to study and look for specific patterns in data for very diverse areas like weather forecasting, customer behavior, events that may directly or indirectly affect oil prices, and so on. When businesses are more aware of specific patterns and trends, they are in a better position to come up with new policies around security and surveillance, or to resolve traffic congestion problems on frequently travelled routes, or even to come up with better hiring policies.

Both government and private organisations stand to benefit tremendously across multiple use cases.

Surveys and reviews

Your text mining API can be programmed to recognise and then categorise specific topics and themes within your, say, customer reviews, market research surveys or emails, using techniques like machine learning, natural language processing, and aspect-based sentiment analysis – ensuring that different aspects and themes present within any piece of information is fully taken into account.

This means that you will no longer have to rely on open-ended and often cryptic responses from customers which can be inaccurate when it comes to analysis, owing to human error and even bias. This way, you can extract better insight from your customer surveys and reviews.

Market research when finding out what customers want or value the most

You’ll find plenty of statistics online which tell that you that customers are indeed very interested in what others have to say about their experiences with a specific business.

According to a Zendesk report, at least 90% of people are influenced by what they read online before making the decision to buy. In fact, they’re even more interested to read what someone has to say about a specific business when the experience or sentiments are negative. Today, with the myriad of review sites that exist, one negative review about your business can really discourage a lot of people from buying your product or service. On the other hand, positive reviews will serve as a kind of social proof, encouraging more and more people to buy from you.

Let’s say you’re about to enter a new market and want to better understand how well your product will be received or whether people will even use it instead of the ones being sold by your competitors.

In this case study, text mining was used to understand what the bestselling speakers on the market were and what kind of features were the most important to consumers when picking up audio equipment.

In the above Amazon case study, the team of analysts wanted to understand what the best kind of speakers were to buy at a price point of $150 – theorising that if you want to bring a new product to market, you must understand the features people value the most. This is actually a great business use case and a fine example of text mining tools at work.

The team extracted data from both structured (ratings, price, etc.) and unstructured (review text) data for five commonly bought speaker brands on Amazon which had received great reviews.

Using consumer ratings, for instance, the data scientists attempted to learn about the characteristics which lead to the best products scores, reviews, and ratings.

Even though it makes sense to automatically choose a product that boast the best sound quality, there are multiple aspects that consumers may consider when buying such a product, like speaker materials, wireless vs. wired, battery life, price range, and so on. And, by reviewing the low-rating topics (all those not related to sound quality), and considering reviewers opinions and sentiments toward those topics, you can learn about the features which are essential for the product you are about to launch.

So, market research done via text mining will help shape your marketing campaign and even expand to other products you’re planning to launch.

Crime prevention and early intervention

With a heinous crime like human trafficking affecting millions of people each year, some brave and extraordinary people have a strong desire to end it. Tom Sabo is one such individual who was deeply disturbed about the harsh realities revolving around human trafficking after attending a symposium.

Tom, working with a skilled team of data analysts, created useful text mining and analytics based models which law enforcement could use to track human trafficking victims. One model, for example, pulled textual data from multiple sources like newspaper articles, police reports, recent prosecutions around human trafficking, and even a shady-looking classified ads website. The main goal was to identify regular patterns in the text so that Tom could integrate that into a predictive analytics model to discover a trail leading him to the victims and perpetrators.

Using police statement from a specific New York jurisdiction, he was able to link the data to events outside the jurisdiction and even those outside the country. Tom identified trends related to not just the victim but also human trafficking events that were currently happening.

His model was subsequently used by law enforcement to identify instances of human trafficking a lot quicker, allowing agencies to take the relevant action before it was too late.

Similarly, you could use text mining to uncover fraud or potentially fraudulent and/or criminal activities in your own organisation or even the industry you are in.

Risk management

Even though we don’t have any concrete examples for this one at the moment, text mining is being actively used in risk management across organisations of all scales. Text mining, in terms of assessing and managing risk, may be used to extract specific information pertaining to industry trends and/or financial markets, through change monitoring in sentiments, or through acquiring information from white papers and analytical reports.

This can be extremely useful for, say, banking and financial institutions because the insights they gain from the above data will allow them to take on investments in different sectors with a lot more confidence. In fact, many banks and financial institutions are taking this approach as we speak.

Product or systems maintenance

Text mining tools can offer an extensive overview of the activity and operation involved in industrial equipment as well as machinery, for example, making it possible for engineers and servicemen to automate maintenance-related decisions.

For instance, text mining can help you discover patterns and trends which might suggest the frequency or occurrence of a specific kind of problem with your machinery. This way, you can implement predictive maintenance measures to intercept issues before they become serious problems and burn a hole in your maintenance budget. As an added bonus, you will also be able to perform maintenance operations more proactively, rather than reactively.

Data analysis on social media

Social media is undeniably one of the ripest sources of unstructured data, and organisations are certainly not turning a blind eye to it. In fact, the sheer amount of insights that can be gathered from social media alone is something that businesses of all scales see as a very valuable market and customer intelligence source.

Text mining APIs can be programmed to help you evaluate or analyse everything from product feedback to brand perception to customer intent, simply by looking at the comments and feedback your customers leave. AI-powered text mining tools and text analytics tools can help to contextualise massive volumes of social commentary – to extract emotions, opinions, and sentiment which reveal precisely what people love (or hate) about your brand, products, and services – really useful information which you can use, in turn, to improve existing products and services or to better tweak upcoming products and/or services.

Spam filtering

Email is still nearly as effective a marketing tool as it was when it was first conceived – a fast, reliable, and relatively cheap way to communicate – but spam has become a major issue for both internet service providers and businesses.

It’s an equally pressing issue for users too as spam serves as an entry point for viruses and phishing, both of which can be severely detrimental to productivity. Text mining techniques, however, can be applied to improve the overall effectiveness of statistical filtering methods once prior knowledge has been fed into the text mining tools.

This makes for safer and far more efficient email management, while also massively improving user experience.

Business intelligence

Highly sophisticated business intelligence tools have now become a mainstay in nearly all enterprise decision-making processes. Text mining tools help to collect, store and analyse data on behalf of your business, allowing you to identify trends, patterns, and opportunities very quickly, compared to manual analysis to discover the above.

With an innovative text mining API at the core, business intelligence tools can better leverage the unstructured data they are fed with, and combined with your own structured data, it can expand your model’s data sets to generate some very valuable business insight.

This, subsequently, can help you create even more advanced business intelligence tools and analysis models which can deliver the critical business insight you need to thrive and move forward.

Better understand your competitors

Whenever your customers leave a comment on your Facebook page, for example, or share specific information when tweeting, it tells you where they are coming from. Similarly, your competitors also leave behind chunks of information whenever they report to public databases. In fact, you can find many instances where the US government has mandated such information, where the data has been made available to the general public.

One company worked with a popular medical device manufacturer while using FDA reports as a data source. The data set contained hundreds to thousands of medical device reports of alleged device-related malfunctions, injuries, and deaths. The company’s team used the data source to connect specific users to various medical device manufacturer’s products. Then they used the text fields to better understand the main issues those users were facing with the some of the medical devices.

A thorough analysis of the unstructured data revealed that the medical device manufacturer had a significantly higher placement success rate compared to their main competitor.

Through topic cluster analysis, a part of data mining, the company demonstrated the most common causes of failed device installations by the manufacturer’s main competitor, which had led to a number of patient deaths. The sophisticated text mining APIs and algorithms provided valuable insights into the manufacturer’s own medical devices and how they stacked up against others in the market, which allowed them to find common problems which could potentially have led to adverse outcomes for many of the devices they were manufacturing. It also offered them insights into what their competitors were doing wrong!

Conclusion

With such an abundance of unstructured text data, organisations have nearly limitless opportunities before them to understand where they stand, where their competitors stand, and how they can better protect themselves from specific risks or act fast when they need to outpace and outsmart the competition.

If you’re not utilising text mining in some form of the other, you are really behind the times and practically handing over a large market share to your competitors on a silver platter. Improve business outcomes and decision making today, among other things, through text mining: +44 20 3488 5740.

Tags

Text mining; text mining tools; text mining API; what is text mining

Meta: Text mining – The many ways businesses can benefit

Your NLP Career Awaits!

Ready to take the next step in your NLP journey? Connect with top employers seeking talent in natural language processing. Discover your dream job!

Find Your Dream Job

Look up company data from names (video)
Ai for business

Look up company data from names (video)

How to look up UK company data from company names (video) Imagine you have a clients list, suppliers list, or investment portfolio…

Unstructured data
Big dataNatural language processing

Unstructured data

Unstructured Data in Healthcare with NLP Introduction In today’s digital healthcare landscape, data plays a pivotal role. However, while medical records, patient feedback, and clinical research generate vast amounts of information, not all of it is easy to manage or analyze.

How to train your own AI: Fine tune an LLM for mental health data
Generative aiAi in research

How to train your own AI: Fine tune an LLM for mental health data

Fine tuning a large language model refers to taking a model that has already been developed, and training it on more data.

What we can do for you

Transform Unstructured Data into Actionable Insights

Contact us