Need Help?

Post UI training resources

Last Updated: Apr 28, 2017
Downloadable guide available as PDF with same information -- see bottom of page.

While the user interface is straightforward, we realize that basic training is a lot to take in. Following is a summary of introductory UI training, as well as a centralized list of starter resources for the newly-minted Luminoso Analyst.
There is an art and science to project analysis. Remember, every data set is unique. Different projects contain different insights, and each data set requires its own analytical approach(es). While discovering major trends and discussions in the data is typically fast and easy, some data sets are more challenging – especially when it comes to less-obvious discussion patterns.

Before starting the insight discovery process and using the tool to identify themes, you need to formulate a hypothesis, at least on a high level.
Quite often the goal of an analytics project is to find root causes of certain consumer or employee emotions and related behaviors so that a business can take appropriate actions in order to correct problems, and enhance services and offerings. Ask yourself common-sense questions about your project with the general knowledge you already possess, such as:
  • What is the data about? Where is it from? Does your data have subsets?
  • What do you expect to find? What known issues are you looking to solve or understand better?
  • Are you looking at trends in customer service emails? Anything specific?
  • Are you trying to find smaller, actionable insights in product reviews for product enhancements?
  • Or are you trying to see what issues may have arisen after a major upgrade in support tickets?
  • Attempting a prediction model about which products have the greatest likelihood of being returned?
  • Are you performing an emotion analysis across subsets (e.g., star ratings) to learn why people assign so many 3-star ratings and not more 5-star ratings?
At a minimum, establish the general direction in which you're headed before performing your analysis. And keep an open mind as you might discover new angles to your hypothesis, which could influence your analysis techniques.
Analysis begins with all-important data discovery and the iterative discovery process learned in basic training (i.e., the Luminoso Analytics Methodology):
  • Examine your top concepts
  • Explore the dynamic concept cloud
  • Drill down into different clusters
  • Isolate concepts in categories (e.g., brand, features, function, services, emotions)
  • Build & edit topics
  • Download analysis exports and examine metrics
  • <repeat>

Start with the basics to familiarize yourself with each data set and then attempt different analysis techniques in order to maximize your analyses and get the most out of your data.

Following are a few highlights of our post-training support materials. Our support site is updated regularly, so make sure to check out these valuable resources in order to help you optimize use of the tool.
QuickStart Guide
The basis for user training, which contains the Luminoso Analytic Methodology, as well as specific information about association scores, topic-subset, and topic-topic analyses.
Theme Identification
Discusses the ability to detect patterns in unstructured text data for grouping, measuring and tracking. It provides instruction on how to isolate themes in your data and formulate a hypothesis before jumping into your analysis.
Association Scores General Info
Association Scores: Analysis and Interpretation Best Practices & Topic-Topic Scores & Best Practices
Know what scores mean! Higher scores indicate higher concept prevalence within a discussion. The bigger the number, the more people are talking about the concept. There are different thresholds according to the source metric (top concepts = higher – vs – topic-topic / topic-subset = lower).
Create a new project from a Topic
Focus technique to evaluate a concept within its own project, in order to reduce some of the "noise" in large data sets and focus on the individual relationships that emerge from this handy analysis technique. Identify a concept with a ~5-15% occurrence of exact matches, download the matching documents, and then upload that CSV into a new project. Provides a new lens for data discovery.
Emotion Analysis & Detecting Emotions in Your Data (video)
This is a major differentiator of Luminoso technology, which captures the full range of complex human emotion to help translate your analysis into specific, actionable insights. Hint: Negative emotions are more nuanced, and they contain a wealth of valuable information (more so than positive emotions).
Differential Analysis (video)
Compare and contrast groups of documents as they evolve over time or across various meta-data fields (NPS, Rating, Demographic).
Predictions on dynamic data (video)
Prediction is the holy grail of marketing. Marketers traditionally use demographic metadata in order to chase their customers' next transaction. With the right combination of Luminoso analysis techniques, you can create long-term customer loyalty by generating more nuanced and targeted insights, which helps marketing see more clearly into the future.

Dashboards & Widgets

Designed as integrated reporting tools to communicate relevant insights to the right people, the Dashboards module provides the ability to visualize, package, and share findings from your data. Download the guide for step-by-step instruction on how to create Dashboards and share them.

Python users can quickly and easily augment their analytics capabilities by leveraging two pre-packaged scripts:
Subset key terms
This script identifies key concepts unique to each subset within a project. Helpful in identifying patterns within subsets that differentiate one group from another (especially helpful with star ratings).
Create Conceptual Subsets from Topics
Once you have built your topics, they can be transformed into subsets. For example, for product reviews of say, a tablet (iPad, Samsung Galaxy, Kindle Fire), create feature-based topics: display, sound, battery life, navigation, camera, connectivity. Run the script and these feature-topics become subsets. Now you can now create descriptor and emotion topics, which can be analyzed across subsets.

More Support
seconds ago
a minute ago
minutes ago
an hour ago
hours ago
a day ago
days ago
Invalid characters found