Natural Language Querying enables users to ask an analytical question in plain English, and the system will respond with the answer in the form of a visualization. The Chat Bot extends this functionality by letting you "chat" in plain English with the system, letting it update and refine your visualization, report, or query according to your given questions or statements.
The Chat Bot offers an intuitive and flexible approach to report-building, enabling you to build whatever queries you want, without clicking any buttons, or performing any drag and drop functions. You can even use it to perform interactive functions like drill, dice, swap, filter, and sort.
Depending on your own preferences, you may find Chat Bot querying to be more intuitive than the alternative methods of mouse-clicking and drag and dropping. In this case, the Chat Bot can encourage you to explore the data model in a more spontaneous way, helping you to discover more interesting and insightful findings.
The NLQ Chat Bot can be used to build queries in Discover, and it can also be used to interact with visualizations in Present at runtime.
- Click here to learn more about NLQ Chat Bot in Present
Note: this feature is available with the Enterprise Edition license only.
Accessing the Chat Bot in Discover
The NLQ chat can be accessed from 2 locations:
- New Discovery: when launching a new Discovery, you can access the Ask a Question feature; from this panel you can begin your Chat Bot.
- Report Ribbon: from the Report Ribbon, click the Ask a Question button to open the Chat Bot panel.
Navigating the NLQ Chat Bot
The Chat Bot is primarily a window that allows users to write a sequence of questions into a text box (black arrow below) and trigger the changes to the currently selected visual. After typing in text, users can hit the ENTER key on their keyboard or click the Go button (green arrow)
Reset
Click the Reset button (yellow arrow above) to reset the chat session and the visual back to its original state.
New Query
Click the New Query button (red arrow above) to restart the chat and the new blank query. All hierarchies and measures will be removed from the visual, allowing you to build a new query from scratch.
Entity Prompt
As you type, you will be prompted with key elements from the data model (blue arrow above). You can select these instead of having to type or guess these items. Use of wildcards can also make accessing these items easier (see below)
Suggestions
Click the Suggestions button (purple arrow above) to see a list of the most popular questions for the current model that match the types of questions you are asking in the session. This can give you a good sense of what sort of information people commonly look for in the model where the selected visual is located. This will not show up if there is not enough question history in the system or there are no matches to your line of questioning.
Chat Bot Querying
The Chat Bot employs almost identical techniques to simple NLQ questions. This includes time intelligence, geographic intelligence, pro-noun intelligence as well as tools for handling syntax, like word stemming and a thesaurus. Users can also drive filtering and sorting.
- Click for more details on writing NLQ questions and NLQ intelligence
"Getting Chatty"
The Chat Bot extends beyond the standard NLQ capabilities by allowing users to ask subsequent questions to adjust the original query (regardless of how it was generated). In this way, the Chat Bot is mostly cumulative, so that you can enter a series of follow up questions and the previous stage will be updated accordingly, without a reset or undo of anything. This allows you to explore your data in an intuitive and spontaneous way, quickly adding, adjusting or and removing items as required.
Importantly, the Chat Bot does not change the selected visualization. It will draw the results of the changes using the currently selected visual, unless a new chat session is triggered.
The following guide provides suggestions on how to "chat" with the "bot".
Additive Effect
If users supply a new question to an existing visual, the question is treated as additive to the existing logic. This applies to every question - rather than "resetting" the visual after each take.
Example of the Additive Effect
For example, asking "Show top 10 promotions by sales" as the first NLQ question will show the following result:
However, if there was an existing query:
And then ask "Show top 10 promotions by sales" in the Chat Bot, it would return the additive result - effectively blending the 2 results together:
Adding, Replacing and Removing Entities
Being able to add, replace or remove content is a key capability needed when an existing query is presented:
- ADD: To add hierarchies and measures, you can simply write "add net profit" to add a measure called "net profit" or "show gender" to add a hierarchy called "gender".
- REMOVE: To remove a hierarchy or measure, write "remove sales", "hide sales" or "remove promotions".
- REPLACE: To replace or swap a hierarchy or measure, write "replace sales with expenses", "replace product category with promotions".
You can also build a composite sentence like "Remove sales and net profit, add cost and promotions"
Changing the Visual
As described above, the Chat Bot uses the currently selected visual for subsequent adjustments. This is unlike a fresh NLQ query, which uses Smart Discovery engine to pick the visualization for you. However, visualizations can be changed.
The normal NLQ allows users to specify their preferred visual as part of the question: "Show sales by gender in a grid" . With the Chat Bot, the same technique can be used to switch visuals. For example "change to pie chart", "Switch to a grid" , "Show as column chart" or you could just write "pie chart", or even just "pie".
Wildcards and Keywords
While typing, the Chat Bot will automatically show a list of hierarchies and measures that match the word currently being typed (after first 2 characters). You can use the wildcard character to show hierarchies and measures in the current model in the event you don't know what to type.
- Show Hierarchies and Measures: * is used to show a list of all measures and hierarchies in the model where the given query was built. This can help you to quickly find hierarchies and measures, which is especially useful if you don't know the exact name of the item you're looking for. It can also help you to spot items that might be of interest.
- Show Hierarchies: *h is used to show a list of all hierarchies in the model.
- Show Measures: *m is used to show a list of all measures in the model.
Restart
The key words reset/ restart/ start over can be used to reset the NLQ Chat session and the query. The same function can be performed by clicking the Restart button in the chat panel.