AnnoMI Counselling Dialogue Annotation
Annotation of motivational interviewing counselling dialogues based on the AnnoMI dataset. Annotators label therapist and client utterances for MI techniques (open questions, reflections, affirmations) and client change talk (sustain talk, change talk), with quality ratings for therapeutic interactions.
Configuration Fileconfig.yaml
# AnnoMI Counselling Dialogue Annotation
# Based on Wu et al., ICASSP 2022
# Paper: https://ieeexplore.ieee.org/document/9746035
# Dataset: https://github.com/uccollab/AnnoMI
#
# Task: Annotate motivational interviewing counselling dialogues.
# Label therapist utterances for MI behavior codes and client utterances
# for change talk / sustain talk categories.
#
# MI Behavior Codes (Therapist):
# - Open Question: Questions that invite elaboration
# - Closed Question: Questions with limited answer options
# - Simple Reflection: Repeating/rephrasing what client said
# - Complex Reflection: Adding meaning, emphasis, or interpretation
# - Affirmation: Acknowledging client's strengths or efforts
# - Giving Information: Providing educational content
# - Advise with Permission: Offering advice with client's agreement
#
# Change Talk Categories (Client):
# - Change Talk: Statements favoring change (desire, ability, reasons, need, commitment)
# - Sustain Talk: Statements favoring status quo
# - Neutral: Neither change talk nor sustain talk
annotation_task_name: "AnnoMI Counselling Dialogue Annotation"
task_dir: "."
data_files:
- sample-data.json
item_properties:
id_key: "id"
text_key: "text"
context_key: dialogue_context
output_annotation_dir: "annotation_output/"
output_annotation_format: "json"
annotation_schemes:
# MI behavior code (for therapist turns) or change talk type (for client turns)
- annotation_type: radio
name: behavior_code
description: "Select the MI behavior code (therapist) or change talk type (client) for this utterance"
labels:
# Therapist codes
- "Open Question"
- "Closed Question"
- "Simple Reflection"
- "Complex Reflection"
- "Affirmation"
- "Giving Information"
- "Advise with Permission"
- "Confront"
- "Direct"
# Client codes
- "Change Talk - Desire"
- "Change Talk - Ability"
- "Change Talk - Reasons"
- "Change Talk - Need"
- "Change Talk - Commitment"
- "Sustain Talk"
- "Neutral/Follow"
keyboard_shortcuts:
"Open Question": "o"
"Closed Question": "c"
"Simple Reflection": "s"
"Complex Reflection": "x"
"Affirmation": "a"
"Change Talk - Desire": "d"
"Change Talk - Commitment": "k"
"Sustain Talk": "u"
"Neutral/Follow": "n"
tooltips:
"Open Question": "Questions that cannot be answered with yes/no and invite the client to elaborate"
"Closed Question": "Questions that can be answered with a short response or yes/no"
"Simple Reflection": "Repeating or slightly rephrasing what the client has said"
"Complex Reflection": "Reflecting with added meaning, emphasis, metaphor, or a new interpretation"
"Affirmation": "Statements that recognize client strengths, efforts, or positive behaviors"
"Giving Information": "Providing factual or educational information without directing behavior"
"Advise with Permission": "Offering suggestions or advice after asking or receiving the client's permission"
"Confront": "Directly challenging or arguing against the client (MI non-adherent)"
"Direct": "Giving orders, commands, or directing the client without permission (MI non-adherent)"
"Change Talk - Desire": "Client expresses wanting to change ('I wish I could...', 'I want to...')"
"Change Talk - Ability": "Client expresses ability to change ('I could...', 'I'm able to...')"
"Change Talk - Reasons": "Client provides reasons for change ('It would be better for my health...')"
"Change Talk - Need": "Client expresses need to change ('I need to...', 'I have to...')"
"Change Talk - Commitment": "Client expresses commitment to change ('I will...', 'I'm going to...')"
"Sustain Talk": "Client argues for maintaining the status quo or against change"
"Neutral/Follow": "Client provides information or follows along without clear change or sustain talk"
# MI techniques observed (multiselect)
- annotation_type: multiselect
name: techniques
description: "Select all MI-consistent or MI-inconsistent techniques observed"
labels:
- "Empathy"
- "Autonomy Support"
- "Collaboration"
- "Evocation"
- "Rolling with Resistance"
- "Developing Discrepancy"
- "Normalizing"
- "Decisional Balance"
- "Elicit-Provide-Elicit"
- "Confrontation (MI-inconsistent)"
- "Unsolicited Advice (MI-inconsistent)"
- "None Notable"
tooltips:
"Empathy": "Speaker demonstrates understanding of the other's perspective"
"Autonomy Support": "Therapist supports the client's autonomy and right to choose"
"Collaboration": "Therapist works with rather than directing the client"
"Evocation": "Therapist draws out the client's own motivations for change"
"Rolling with Resistance": "Therapist avoids arguing and reframes resistance"
"Developing Discrepancy": "Therapist helps client see gap between current behavior and goals"
"Normalizing": "Therapist normalizes the client's experience or ambivalence"
"Decisional Balance": "Exploring pros and cons of change and status quo"
"Elicit-Provide-Elicit": "Asking what client knows, sharing information, then asking for reaction"
"Confrontation (MI-inconsistent)": "Therapist directly confronts or argues with client"
"Unsolicited Advice (MI-inconsistent)": "Therapist gives advice without permission"
"None Notable": "No specific MI techniques observed in this utterance"
# Quality rating
- annotation_type: likert
name: quality
description: "Rate the MI quality of this utterance (1=poor MI adherence, 5=excellent MI adherence)"
size: 5
min_label: "1 - Poor MI adherence"
max_label: "5 - Excellent MI adherence"
labels:
- "1 - MI non-adherent: confrontational, directive, or dismissive"
- "2 - Below average: minimal MI skill demonstrated"
- "3 - Adequate: basic MI skills present but room for improvement"
- "4 - Good: clear MI-consistent behavior with minor gaps"
- "5 - Excellent: exemplary MI practice, skillful and empathic"
keyboard_shortcuts:
"1 - MI non-adherent: confrontational, directive, or dismissive": "1"
"2 - Below average: minimal MI skill demonstrated": "2"
"3 - Adequate: basic MI skills present but room for improvement": "3"
"4 - Good: clear MI-consistent behavior with minor gaps": "4"
"5 - Excellent: exemplary MI practice, skillful and empathic": "5"
allow_all_users: true
instances_per_annotator: 100
annotation_per_instance: 2
allow_skip: true
skip_reason_required: false
Sample Datasample-data.json
[
{
"id": "annomi_001",
"text": "Tell me a little about what brings you in today and what's been going on with your drinking.",
"speaker": "therapist",
"dialogue_context": "[Start of counselling session]"
},
{
"id": "annomi_002",
"text": "Well, my wife has been on my case about it. She thinks I drink too much, but honestly I don't think it's that big of a deal. I have a couple of beers after work to unwind, maybe more on weekends.",
"speaker": "client",
"dialogue_context": "Therapist: Tell me a little about what brings you in today and what's been going on with your drinking."
}
]
// ... and 8 more itemsGet This Design
Clone or download from the repository
Quick start:
git clone https://github.com/davidjurgens/potato-showcase.git cd potato-showcase/text/dialogue/annomi-counseling-dialogue potato start config.yaml
Details
Annotation Types
Domain
Use Cases
Tags
Found an issue or want to improve this design?
Open an IssueRelated Designs
Clickbait Detection (Webis Clickbait Corpus)
Classify headlines and social media posts as clickbait or non-clickbait based on the Webis Clickbait Corpus. Identify manipulative content designed to attract clicks through sensationalism, curiosity gaps, or misleading framing.
Conversation Quality Attributes
Dialogue quality assessment based on controllable dialogue generation research (See et al., NAACL 2019). Annotators evaluate conversation turns for engagement quality, rate overall conversation quality, and identify specific dialogue attributes.
Deceptive Review Detection
Distinguish between truthful and deceptive (fake) reviews. Based on Ott et al., ACL 2011. Identify fake reviews written to deceive vs genuine customer experiences.