Structured Summary Upgrade for AutoSummary

​

Feature Release

This is an announcement for an upcoming ASAPP feature. Your ASAPP account team will provide a target release date and can direct you to more detailed information as needed.

​

Overview

With this upgrade, ASAPP is enhancing the structured summary tags feature of our AutoSummary product. Structured summary tags provide conversation summaries formatted as a set of analytics-ready descriptive tags, with each tag meant to represent a key action from the conversation.

We are updating structured summary tags with the following new capabilities:

• A significant upgrade in the level of granularity presented in each summary
• A custom label space, based on historical conversations (instead of industry-based)
• The inclusion of key customer-specific nomenclature (e.g. product names) in the returned tags

​

Use and Impact

This upgrade to AutoSummary’s structured summary tags supports the same analytics and reporting-related use cases as the existing feature. The higher levels granularity and detail can enable customers to discover more actionable and impactful insights. In particular:

• More granular detail enables more fine-grained data filtering and slicing and dicing
• ASAPP has proven that the enhanced summaries provided by this upgrade has higher predictive power than the current version, useful for finding stronger and more descriptive correlations with business KPIs
• This enhancement’s higher descriptive power is helpful for finding automation and self-serve flows and opportunities

​

How It Works

The following table compares the upgrade to the previous version of structured summary tags:

The following table shows the definition of each component of the new ontology:

​

API Response

This upgrade comes with no changes to the API specs; requesting structured summary tags will continue as before, with a minor update to the provided response:

• The content of the “topic” & “act” fields will be more detailed, and can be customer-specific
• An additional field called “modifier” is provided, adding additional detail to the topic and increasing the level of granularity of Structured Summary Tags

​

FAQs

1. How do you define the label space for the summary tags?

   We use an ML-based technique to reflect and standardize the steps the agent took to address the issue(s) at-hand and other important actions by either party in the conversation, making explicit the result of the conversation.

   We then use other ML-powered heuristics to transform those standardized key actions and events into the “actor-action-topic-modifier” ontology.

2. How do you keep the label space up to date?

   Every month, ASAPP checks if a new relevant tag should be added to the label space. This happens if a relevant action or event can’t be represented by the tags that already exist in the label space. ASAPP expects in the future to increase the update frequency to a point in which tags are added into the label space automatically as soon as there’s enough statistical evidence to conclude they correspond to a new relevant action or event.

3. What’s the relationship between free-text summaries and structured summary tags?

   Structured summary tags can be understood as the structured and standardized representation of those sentences in the free-text summary that summarize the resolution process and result of the conversation.

4. What’s the relationship between structured summary tags and intents?

   Structured summary tags are focused on portraying the resolution process and result of the conversation. On the other hand, the intent (which is requested in a different endpoint) is focused exclusively on the contact reason.

   When combining intents with structured summary tags you end up getting a structured blueprint of the conversation, from start to end.