Chart building user interface providing machine learned chart recommendations
Systems and methods are disclosed to implement a chart recommendation system that recommends charts to users during a chart building process. In embodiments, when a new chart is being created, specified features of the chart are provided to a machine learned model such as a self-organizing map. The model will determine a previous chart that is the most similar to the new chart and recommend the previous chart to the user for recreation. In embodiments, newly created charts are added to a library and used to update the model. Charts that are highly popular or authored by expert users may be weighed more heavily during model updates, so that the model will be more influenced by these charts. Advantageously, the disclosed system allows novice users to easily find similar charts created by other users. Additionally, the disclosed system is able to automatically group similar charts without using human-defined classification rules.
1. A system, comprising:
one or more hardware processors with associated memory that implement a chart recommendation system, configured to:
receive, via a graphical user interface (GUI), user input to create a chart indicating features of the chart;
encode the features of the chart into a feature vector;
provide the feature vector as input to a machine learned model, wherein the machine learned model encodes a chart search structure that organizes a plurality of previous charts into sets of one or more similar charts according to a similarity metric calculated based on respective features vectors of the previous charts;
determine, using the machine learned model, one or more similar charts from the plurality of previous charts that are most similar to the chart according to the similarity metric;
display, via the GUI, the one or more similar charts as one or more recommended charts using respective feature vectors of the one or more similar charts;
receive, via the GUI, a user selection of one of the one or more recommended charts;
save the recommended chart selected as a new chart in a database;
update a popularity metric of the recommended chart selected, wherein the popularity metric is displayed with the recommended chart via the GUI; and
retrain the machine learned model based on the popularity metric.
2. The system of claim 1 , wherein
the chart recommendation system is configured to:
save the new chart in a chart library used for future chart recommendations; and
update the machine learned model using features of the new chart to enable the new chart to be included in searches for the future chart recommendations.
3. The system of claim 1 , wherein
the machine learned model is a neural network.
4. The system of claim 1 , wherein
the machine learned model is a self-organizing map (SOM).
5. The system of claim 1 , wherein
the recommended chart is created by a first user; and
the new chart is created by a second user and associated with the second user in the database.
6. The system of claim 1 , wherein the feature vector encodes one or more of:
a name of the chart;
a name of a data source that stores data used in the chart;
a query used to retrieve the data from the data source;
a time length specified for the chart; and
a chart type of the chart.
7. The system of claim 1 , wherein
the chart recommendation system is configured to output a recommendation indicating a size of the recommended chart or a position of the recommended chart in a layout group.
8. The system of claim 1 , wherein
the chart recommendation system is configured to change recommended charts in response to user modifications to the chart received via the GUI.
9. The system of claim 1 , wherein
the chart recommendation system is configured to periodically retrain the machine learned model using additional charts saved to the database.
10. The system of claim 1 , wherein
the chart recommendation system is configured to:
create a custom machine learned model for a particular user group; and
use the custom machine learned model to generate recommendations for additional charts created by users in the user group.
11. The system of claim 1 , wherein
the chart recommendation system is configured to:
create a custom machine learned model for a particular type of report; and
use the custom machine learned model to generate recommendations for additional charts created for the particular type of report.
12. The system of claim 1 , wherein
the chart recommendation system is configured to:
display, via the GUI, a list of recommended charts for the chart, wherein the list of recommended charts is ranked based on the similarity metric of the recommended charts or a popularity score of the recommended charts.
13. The system of claim 1 , wherein
the GUI is a web-based interface displayed via a web browser.
14. The system of claim 4 , wherein
the similarity metric is a Euclidean distance between feature vectors.
15. The system of claim 4 , wherein
to determine the one or more similar charts, the chart recommendation system is configured to:
perform an initial search to determine a best matching unit (BMU) in the SOM based on the feature vector;
determine that an insufficient number of charts are mapped to the BMU; and
perform an expanded search for similar charts in neighboring units of the BMU.
16. The system of claim 6 , wherein
the feature vector encodes one or more of:
a name of a layout group specified for the chart, wherein charts in the layout group are displayed together;
one or more names of one or more other charts in the layout group;
a position of the chart in the layout group; and
a width and height of the chart in the layout group.
17. The system of claim 9 , wherein
the chart recommendation system is configured to retrain the machine learned model based on a user attribute of users who created the additional charts.
18. A method comprising:
performing, by one or more hardware processors with associated memory that implement a chart recommendation system:
receive, via a graphical user interface (GUI), user input to create a chart indicating features of the chart;
encode the features of the chart into a feature vector;
provide the feature vector as input to a machine learned model, wherein the machine learned model encodes a chart search structure that organizes a plurality of previous charts into sets of one or more similar charts according to a similarity metric calculated based on respective features vectors of the previous charts;
determine, using the machine learned model, one or more similar charts from the plurality of previous charts that are most similar to the chart according to the similarity metric;
display, via the GUI, the one or more similar charts as one or more recommended charts using respective feature vectors of the one or more similar charts;
receive, via the GUI, a user selection of one of the one or more recommended charts;
save the recommended chart selected as a new chart in a database; and
retrain the machine learned model using an additional chart, wherein the machine learned model is retrained based on a user attribute of a user who created the additional chart.