Apply suggestions from code review

Acrolinx fixes

Author: 19BMG00

articles/cognitive-services/language-service/conversational-language-understanding/how-to/migrate-from-luis.md

@@ -19,7 +19,7 @@ ms.custom: language-service-clu, ignite-fall-2021
 
 CLU offers the following advantages over LUIS: 
 
-- Improved accuracy with state-of-the-art machine learning models for better intent classification and entity extraction. LUIS required a more examples to generalize certain concepts in intents and entities, while CLU's more advanced machine learning reduces the burden on customers by requiring significantly less data.  
+- Improved accuracy with state-of-the-art machine learning models for better intent classification and entity extraction. LUIS required more examples to generalize certain concepts in intents and entities, while CLU's more advanced machine learning reduces the burden on customers by requiring significantly less data.  
 - Multilingual support for model learning and training. Train projects in one language and immediately predict intents and entities across 96 languages.
 - Ease of integration with different CLU and [custom question answering](../../question-answering/overview.md) projects using [orchestration workflow](../../orchestration-workflow/overview.md). 
 - The ability to add testing data within the experience using Language Studio and APIs for model performance evaluation prior to deployment. 
@@ -207,10 +207,10 @@ There are several features that were present in LUIS that will no longer be avai
 
 - **Patterns**: Patterns were added in LUIS to assist the intent classification through defining regular expression template utterances. This included the ability to define Pattern only intents (without utterance examples). CLU is capable of generalizing by leveraging the state-of-the-art models. You can provide a few utterances to that matched a specific pattern to the intent in CLU, and it will likely classify the different patterns as the top intent without the need of the pattern template utterance. This simplifies the requirement to formulate these patterns, which was limited in LUIS, and provides a better intent classification experience. 
 
-- **Phrase list features**: The ability to associate features mainly occured to assist the classification of intents by highlighting the key elements/features to use. This is no longer required since the deep models used in CLU already possess the ability to identify the elements that are inherent in the language. In turn removing these features will have no effect on the classification ability of the model.
+- **Phrase list features**: The ability to associate features mainly occurred to assist the classification of intents by highlighting the key elements/features to use. This is no longer required since the deep models used in CLU already possess the ability to identify the elements that are inherent in the language. In turn removing these features will have no effect on the classification ability of the model.
 
 - **Structured entities**: The ability to define structured entities was mainly to enable multilevel parsing of utterances. With the different possibilities of the sub-entities, LUIS needed all the different combinations of entities to be defined and presented to the model as examples. In CLU, these structured entities are no longer supported, since overlapping learned components are not supported. There are a few possible approaches to handling these structured extractions:
-    - **Non-ambiguous extractions**: In most cases the detection of the leaf entities is enough to understand the required items within a full span. For example, structured entity such as _Trip_ that fully spanned a source and destination (_London to New York_ or _Home to work_) can be identified with the indivdual spans predicted for source and destination. Their presence as individual predictions would inform you of the _Trip_ entity.
+    - **Non-ambiguous extractions**: In most cases the detection of the leaf entities is enough to understand the required items within a full span. For example, structured entity such as _Trip_ that fully spanned a source and destination (_London to New York_ or _Home to work_) can be identified with the individual spans predicted for source and destination. Their presence as individual predictions would inform you of the _Trip_ entity.
     - **Ambiguous extractions**: When the boundaries of different sub-entities are not very clear. To illustrate, take the example “I want to order a pepperoni pizza and an extra cheese vegetarian pizza”. While the different pizza types as well as the topping modifications can be extracted, having them extracted without context would have a degree of ambiguity of where the extra cheese is added. In this case the extent of the span is context based and would require ML to determine this. For ambiguous extractions you can use one of the following approaches:
 
 1. Combine sub-entities into different entity components within the same entity.