The challenge for insight professionals lies in the overwhelming volume of information proliferated through news sources, social media, and various reports. Gaining an overview often takes excessive amounts of time, and the large volumes make it difficult to discern the unique insights. Many texts are also inaccessible to researchers due to language barriers.
Dcipher’s landscaping workflows can give you a quick overview of large volumes of text in the form of an interactive “landscape map” that surfaces important information, including “unknown unknowns” that you didn't know existed. To delve deeper, Dcipher Research Bots answer your questions about everything you want to understand in detail and provide references to the original sources for validation and further reading.
Typical use cases
Speeding up all kinds of desk research and information gathering processing
Mapping topics, themes, and trends in news reporting and social media conversations around topics of interest
Getting answers to questions and finding the most relevant sources for topics of interest
Delivery formats and outputs
An interactive report with a content landscape based on news reporting, social media, or PDF reports in your area of interest:
And a Dcipher Research Bot that you can chat with to get answers and summaries based on the news articles, social media posts, and reports you that you have asked it to read:
How to utilize Dcipher Analytics’ Insight Booster Toolkit?
The overall approach described here has two components:
Divergent phase: first an open-ended, bottom-up content landscaping analysis of an area of interest to get an overview and surface interesting topics and themes.
Convergent phase: then a top-down analysis, interacting with a research bot to delve deeper into topics discovered in the divergent phase.
Each of the two steps can be run separately. Content landscaping is especially relevant when trying to understand an area and for surfacing unexpected patterns and information. Research bots are most useful when the questions are known, and when finding the answers to the questions is like searching for needles in a haystack.
Step 1: Run a content landscaping workflow
Log into your Dcipher Analytics account, go to “Dcipher Workflows”, and select one of the following workflows for content landscaping:
Click “Use this workflow” and follow the instructions in the workflow wizard. The wizard guides you through the process of setting up the workflow. At the end of the Wizard, click “Run workflow” to start the workflow run and wait for the results to be generated. You will get notified by email once your delivery is ready.
Step 2: Access the results via an interactive report
Once the workflow is done, click the workflow in the Workflow view and then click the interactive report icon. This will open the interactive report so you can view and interact with the results. The report has three views:
Content Landscape: Shows topics (clusters of similar information) and thematic hotspots (clusters of topics) organized based on similarity. Click a topic to get a summary of the associated information and key source references. This view is a great way for getting an overview of the analyzed content and finding interesting topics to delve deeper into.
Category View: Shows topics organized by available categories, such as themes and countries. This view is useful for dividing the identified topics into different categories, making them easier to digest.
Trend Chart: Shows the growth in the topics’ share of voice over different time periods, depending on what the input data supports. This view provides a great way of finding topics that are trending over a given time period.
To publish an interactive report publicly for sharing with others, click the “Copy link” icon in the top-right corner of the screen.
Exploring an interactive report often gives rise to new questions and an appetite for drilling deeper into topics of interest. This is where the research bot comes in (see subsequent steps).
Step 3: Set up a research bot to train on your content or external data
When logged into your Dcipher Analytics account, go to “Dcipher Workflows”, and select one of the following workflows for training a research bot depending on the source of interest:
After clicking “Use this workflow”, the wizard guides you through the process of training the research bot. When the training is done, you’ll get an email notification.
Step 4: Interact with the research bot
You find the trained research bot on the Research Bots page in your Dcipher Analytics account. Click the bot to start interacting with it. There are three interaction modes:
“Question-answer” is the best mode when you have concrete questions that you want an answer to. Ask anything you want to know related to the topic that you’ve set up the research bot to read up on.
“Subtopic summarization” performs a content analysis and summarizes the seven largest themes within the data.
The bot will let you know if there is no relevant information about your question or topic in the textual data it has been set up to read. In that case you may consider expanding the search space and feeding it with more relevant information.
Use the filter icon to get answers or summaries based on a subset of the full data, for example news reporting from a specific country or information from a specific report.
Example analysis
Running a content landscaping workflow (step 1):
Exploring the interactive report (step 2):
Training a research bot (step 3):
Interacting with the research bot (step 4):