Merge main into dev May 25th

CITATION.cff

@@ -0,0 +1,24 @@
+cff-version: 1.0.0
+message: "If you use Azimuth in your projects, please cite it as below."
+authors:
+- family-names: "Branchaud-Charron"
+  given-names: "Frederic"
+- family-names: "Gauthier-Melancon"
+  given-names: "Gabrielle"
+- family-names: "Marinier"
+  given-names: "Joseph"
+- family-names: "Brin"
+  given-names: "Lindsay"
+- family-names: "Tyler"
+  given-names: "Chris"
+- family-names: "Le"
+  given-names: "Di"
+- family-names: "Grande"
+  given-names: "Karine"
+- family-names: "Babu"
+  given-names: "Nandhini"
+title: "Azimuth, an open-source dataset and error analysis tool for text classification"
+version: 2.0.0
+doi: 10.5281/zenodo.6511558
+date-released: 2022-04-20
+url: "https://github.com/ServiceNow/azimuth"

README.md

@@ -111,28 +111,3 @@ add other `README` files in appropriate places, and don't forget to edit them if
 ## License
 
 The package is licensed by ServiceNow, Inc. under the Apache 2.0 license. See [LICENSE](LICENSE) for more details.
-
-## Citation
-
-If you have used Azimuth in your projects, you can use the following citation:
-
-```
-@software{azimuth_2022,
-  author    = {Branchaud-Charron, Frederic and
-               Gauthier-Melancon, Gabrielle and
-               Marinier, Joseph and
-               Brin, Lindsay and
-               Tyler, Chris and
-               Le, Di and
-               Grande, Karine and
-               Babu, Nandhini},
-  title     = {{Azimuth, an open-source dataset and error analysis
-                tool for text classification}},
-  month     = may,
-  year      = 2022,
-  publisher = {Zenodo},
-  version   = {2.0.0},
-  doi       = {10.5281/zenodo.6511558},
-  url       = {https://doi.org/10.5281/zenodo.6511558}
-}
-```

azimuth/modules/model_performance/outcomes.py

@@ -68,7 +68,7 @@ def compute_on_dataset_split(self) -> List[OutcomeResponse]:  # type: ignore
         """
         dm = self.get_dataset_split_manager()
         ds = assert_not_none(self.get_dataset_split())
-        labels = ds["label"]
+        labels = ds[self.config.columns.label]
 
         model_predictions = self._get_predictions(without_postprocessing=True)
         model_outcomes: List[OutcomeName] = [

docs/docs/about-us.md

@@ -110,4 +110,4 @@ Thank you to all others who helped us developing this!
   contributions and continuous feedback.
 * Nikola Simic for creating the visual identity of Azimuth.
 * Michael Lanoie for feedback on our documentation.
-* Sean Hugues, Sethu Meiyappan and Francis Ruel for helping us on the path to open sourcing.
+* Sean Hughes, Sethu Meiyappan, and Francis Ruel for helping us on the path to open sourcing.

docs/docs/development/dev-practices.md

@@ -1,12 +1,18 @@
 # Development Pratices
 
-## New Branch
-When starting a new branch, we don't have a convention for the name.
-You can use your initials or your surname, and/or include the issue number according to your preference.
-It should at the very least include a short description of the issue.
+## Git flow
 
-!!! example "Branch name examples"
-    `pr/fred/fixing-poetry`, `AZ-345/fixing-poetry`, `ggm/fixing-poetry`...
+We follow an approach named [Git flow](https://www.atlassian.com/git/tutorials/comparing-workflows/gitflow-workflow):
+
+* `main`: Should always be working with up-to-date documentation, users will run this branch.
+* `dev`: Integrates new features, documentation can be lacking.
+* `feature/*`: Feature branch than can be merged into `dev` only.
+* `hotfix/*`: Hotfix branches that can be merged into `main`.
+* `release/*`: Not used for now.
+
+Each version has its own tag (:material-tag:`v2.0.0` for example) and release package on Github.
+
+When hotfixes are merged into `main`, they should be ported back into `dev` as well.
 
 ## Committing
 We try to commit regularly, so it is easy to revert partial changes.

docs/docs/user-guide/exploration-space/index.md

@@ -69,7 +69,7 @@ filters are listed below.
 
 ![Screenshot](../../_static/images/control-panel/utterances-search.png){: style="width:200px"}
 
-* **Search a particular string** to filter utterances that contains it.
+* **Search a particular string** to filter utterances that contain it.
 * Filter predictions based on their **confidence value**. You can specify a minimum and a maximum
   value.
 * Filter predictions according to their prediction [**outcomes**](../../key-concepts/outcomes.md).

docs/docs/user-guide/exploration-space/performance-overview.md

@@ -46,6 +46,9 @@ while incorrect predictions include `Incorrect & Predicted` and `Incorrect & Rej
 clouds change depending on the filters and indicate the frequency of each important word across all
 filtered utterances.
 
+Clicking on a word will filter utterances that contain it. Clicking it a second
+time will clear that search filter, so does the ✗ in the `Search utterances` filter.
+
 ![Screenshot](../../_static/images/exploration-space/word-cloud.png){: style="width:400px"}
 
 !!! note "What does the word cloud count?"

webapp/src/components/Analysis/UtterancesTable.tsx

@@ -349,7 +349,7 @@ const UtterancesTable: React.FC<Props> = ({
   const searchString = constructSearchString(pipeline);
   const RowLink = (props: RowProps<Row>) => (
     <Link
-      style={{ color: "unset" }}
+      style={{ color: "unset", textDecoration: "unset" }}
       to={`/${jobId}/dataset_splits/${datasetSplitName}/utterances/${props.row.id}${searchString}`}
     >
       <GridRow {...props} />

webapp/src/components/TopWords/TopWords.tsx

@@ -80,34 +80,33 @@ const TopWords: React.FC<Props> = ({
             No word count data.
           </Typography>
         ) : (
-          wordCounts.map((wordCount, index) => (
+          wordCounts.map(({ word, count }, index) => (
             <Link
               key={index}
               component={RouterLink}
               to={`${baseUrl}${constructSearchString({
                 ...filters,
-                utterance: wordCount.word,
+                utterance: filters.utterance === word ? undefined : word,
                 ...pagination,
                 ...pipeline,
                 ...postprocessing,
               })}`}
               color={theme.palette[palette].dark}
               display="block"
               fontSize={`${Math.max(
-                maxFontSizeInEm * wordCount.count * normalizingScaleFactor,
+                maxFontSizeInEm * count * normalizingScaleFactor,
                 minFontSizeInEm
               )}em`}
               marginLeft={1}
               marginRight={1}
               sx={{
-                opacity: Math.max(
-                  wordCount.count * normalizingScaleFactor,
-                  minOpacity
-                ),
+                opacity: Math.max(count * normalizingScaleFactor, minOpacity),
+                textDecorationStyle:
+                  filters.utterance === word ? "double" : undefined,
               }}
               whiteSpace="nowrap"
             >
-              {`${wordCount.word} (${wordCount.count})`}
+              {`${word} (${count})`}
             </Link>
           ))
         )}

webapp/src/components/Utterance/SimilarUtterances.tsx

@@ -123,7 +123,7 @@ const SimilarUtterances: React.FC<Props> = ({
   const searchString = constructSearchString(pipeline);
   const RowLink = (props: RowProps<Row>) => (
     <Link
-      style={{ color: "unset" }}
+      style={{ color: "unset", textDecoration: "unset" }}
       to={`${baseUrl}/${props.row.id}${searchString}`}
     >
       <GridRow {...props} />