Pith. sign in

REVIEW 5 major objections 6 minor 76 references

Reframe Your Life Story: Interactive Narrative Therapist and Innovative Moment Assessment with Large Language Models

T0 review · 5 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper argues that an LLM can act as a narrative therapist by explicitly planning therapeutic stages and reflection levels, and that doing so produces measurably more transformative dialogue than generic role-playing.

desk verdict Novel theory-grounded LLM therapy framework, but the IMA salience metric is confounded by therapist response length and the paper overstates its main quantitative claim. read the letter →

arxiv 2507.20241 v2 pith:YZ34IYGQ submitted 2025-07-27 cs.CL

Yi Feng , Jiaqi Wang , Wenxuan Zhang , Zhuang Chen , Yutong Shen , Xiyao Xiao , Minlie Huang , Liping Jing
show 1 more author
Jian Yu
This is my paper · ORCID
classification cs.CL
keywords narrativetherapyinteractivetherapistinnovativemomentslargelanguagemodelstherapeuticstageplanningretrieval-augmentedgenerationdialogueevaluationmentalhealthsupport
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper sets out to show that an LLM can simulate a narrative therapist the way the therapy is actually practiced: moving through ordered stages—trust building, problem externalization, re-authoring, and re-membering—and deliberately adjusting how deeply it probes at each step. The authors claim this staged, progression-aware approach makes simulated therapy measurably more effective than asking a generic LLM to role-play a therapist. They support the claim with evaluations over 260 simulated clients and 230 human participants, and with a new assessment tool, IMA, that counts 'innovative moments'—client statements that break with a problem-saturated story—and tracks them across a session. If correct, the work matters because it turns an empirically grounded theory of change into a computational protocol that can be deployed, evaluated, and improved without requiring a human therapist in the loop.

What carries the argument

The load-bearing object is the planning space $\Phi=(\mathcal{S},\mathcal{L})$, a formal encoding of narrative therapy's structure: $\mathcal{S}$ is the set of four therapeutic stages and $\mathcal{L}$ is the set of reflection depths available inside each stage. Two prompt-defined functions, $\Psi_S$ and $\Psi_L$, read the dialogue history and the current client utterance to pick the next stage and the next reflection level, and a third function $\Psi_T$ generates the response conditioned on that state plus $k=5$ retrieved expert exemplars. The companion evaluation machinery is the IMA classifier $\Gamma$, which labels each client utterance with one or more of six innovative-moment types (Action, Reflection, and Protest, at two levels), and the salience formula of Equation (5), which converts those labels into a session-level measure of narrative change. The machinery works together to test the paper's hypothesis: that state planning produces the depth and pacing that role-playing instructions alone do not.

What would settle it

Have two expert narrative therapists annotate the intended therapeutic stage and reflection depth for every turn of the collected dialogues, then compare their labels with INT's planned states; if agreement is near chance, and if an ablation that replaces the planned state with a random state while keeping the same retrieved exemplars matches full INT's IM salience and dimension scores, the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that explicit planning of the therapeutic state is what separates a transformative LLM therapist from a merely supportive one. INT maintains a planning space of four stages and multiple reflection levels per stage; before each reply it uses prompt-driven functions to choose the next stage and reflection depth, then retrieves expert-style exemplars for that state and generates a short, conversational response. IMA operationalizes the clinical construct of innovative moments by classifying each client utterance into six types at two levels and computing their salience as the fraction of dialogue words that carry narrative change. The paper's evidence shows INT beating direct role-playing baselines on the four narrative-therapy dimensions in human evaluation, reaching the highest overall IM salience (29.7% versus a next-best 21.9%) and the highest Level 2 markers, and the same pattern holds in the automated simulated-client study. The authors also report that removing either retrieval augmentation or reflection-level planning degrades performance, with the full system advancing further into the re-authoring and re-membering stages than its ablations.

Load-bearing premise

The load-bearing premise is that the stage and reflection level the planner picks are accurate and informative; the paper never measures those labels against expert judgment, so the measured advantage could in principle come from the retrieved exemplars or the response style rather than from progression-aware planning.

Editorial extensions

If this is right

  • A process-aware therapy simulator can be built by prompting a general LLM with a stage-and-reflection planner, without task-specific fine-tuning for the therapist role.
  • The IMA salience metric offers a way to compare therapy systems on clinically meaningful narrative change rather than on static empathy ratings or surface overlap metrics.
  • Because INT can synthesize support conversations, the resulting NTConv dialogues can be used to fine-tune smaller models; the paper reports that a model trained on them outperforms one trained on the original ESConv dialogues on both automatic and human metrics.
  • The observed Level 1-to-Level 2 innovative-moment shift across turns provides process-level evidence that structured planning induces the kind of narrative progression seen in successful therapy, not merely compliant client agreement.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An unstated corollary is that the same planner could be turned into a supervision display for human therapists, showing where a session sits in stage and reflection depth; the paper does not test this, but its planning space is exactly the information such a tool would need.
  • The IMA trajectory—Level 1 moments early, Level 2 moments late—could be used as a training signal: a therapist optimized to elicit Level 2 innovative moments would, if the paper's theory is right, produce better therapeutic outcomes than one optimized for empathy alone.
  • Because the stage and reflection labels come from the same model class that generates the responses, the causal role of planning is not fully separated from retrieval and style; a cleaner test would freeze the planner and vary only the response generator, or vary the planner while holding the responder fixed.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper proposes INT, a planning-based LLM system that simulates narrative therapists by tracking therapeutic stages and reflection levels and using retrieval-augmented generation, together with IMA, an evaluation method that operationalizes the Innovative Moments Coding System to measure narrative transformation in client speech. The authors report automated experiments with 260 simulated clients and human experiments with 200 main plus 30 follow-up participants, claiming that INT outperforms standard role-playing LLMs across therapeutic dimensions and nearly doubles the elicitation of advanced (Level 2) innovative moments. They additionally fine-tune Qwen3-8B on an INT-generated NTConv dataset and report improvements over ESConv on automatic and human metrics.

Significance. If the central claims hold, this is a useful step toward theory-grounded, process-aware therapeutic dialogue systems and toward automatic, therapy-grounded evaluation. The strengths of the paper are the external grounding of IMA in the IMCS coding system, the use of human expert annotators with reported inter-rater reliability (Cohen's kappa > 0.75), the relatively large human study (230 participants), and the planned release of code and data. However, the headline quantitative claim that INT 'nearly doubles' advanced innovative moments is currently confounded by the definition of the salience metric, which includes therapist words in the denominator, and the automated evaluation is partly self-referential because GPT-4o plays the client, the therapist, and the supervisor. These issues make the magnitude of the reported improvements uncertain rather than definitively wrong, so the paper merits major revision.

major comments (5)
  1. [§4.1.2, §4.1.3, Table 2] The IMA salience denominator includes therapist words: Salience(I_i) = Σ WordCount(C_t ∩ I_i) / Σ WordCount(C_t ∪ T_t). Because INT produces deliberately shorter therapist turns (66.1 vs. 113.0 words per turn in Fig. 4a), the reported Level-2 SUM is inflated. For the human evaluation, INT's Level-2 SUM is 19.408% and GPT-4o's is 10.826%, a ratio of about 1.79, which underlies the 'nearly doubling' claim. Using the average turn lengths in Fig. 4(a), the approximate IM word counts per turn are 0.19408 × (38.8 + 66.1) ≈ 20.4 for INT and 0.10826 × (31.5 + 113.0) ≈ 15.6 for GPT-4o, a ratio of about 1.30; with a client-only denominator the salience figures become approximately 52.5% and 49.7%, near parity. The paper does not report absolute IM word counts or client-only salience, so the central demonstration of INT's superiority in eliciting advanced narrative markers is not yet supported. Please reanalyze the IMA results with client-only denominators or absolute counts, and temper the abstract claim accordingly.
  2. [§4.2, Table 3] The automated evaluation is partly self-referential: GPT-4o simulates the client, GPT-4o implements INT's planning and response generation, GPT-4o acts as the counseling supervisor for dimension scoring, and GPT-4o performs IM annotation. Since INT is built on GPT-4o prompts, the automated comparison may reward responses that match GPT-4o's own stylistic preferences rather than genuine therapeutic quality. The paper reports no agreement study between the GPT-4o supervisor/annotator and human experts on the same automated dialogues. I recommend reporting human--LLM annotation agreement on a shared sample and varying the client and supervisor models to ensure the Table 2 conclusions are not an artifact of model self-preference.
  3. [§4.1.3, Table 3] The claim in the abstract and §4.2 that INT 'consistently outperforms standard LLMs in therapeutic quality and depth' is overstated: in the human evaluation, INT's Reassuring score is 3.09, below GPT-4o's 3.11 and Claude-3.7-sonnet's 3.08. The paper reports no error bars or significance tests for any of the human-rated dimensions, so the remaining differences, while directionally favorable to INT on Empowering, Transformative, and Reconnecting, are not shown to be statistically reliable. Please report per-participant variance, confidence intervals, and appropriate significance tests, or qualify the claims.
  4. [§3.2.2, Eqs. (1)-(2), Fig. 4(b)] The human expert IM annotations cover only 10 randomly sampled sessions per system. The IMA percentages in Table 3 are therefore based on n = 10 per condition, and no per-session variance or confidence intervals are reported. Given the high variability of client narratives, a difference in Level-2 SUM of 19.4% versus 10.8% could be driven by a few unusual sessions. Please provide per-session distributions, bootstrap confidence intervals, or a mixed-effects analysis for the IMA comparisons.
  5. [§4.2] The accuracy of the stage planner Ψ_S and reflection-level planner Ψ_L is never measured against expert judgment. The paper reports stage and reflection-level distributions in Fig. 4(b), but not whether these inferred states match what a trained narrative therapist would choose. Because the ablations (w/o RAG and w/o RAGRL) change both the planning signal and the response-generation prompt, the observed improvement cannot be unambiguously attributed to progression-aware planning. I recommend evaluating planning accuracy on a labeled subset, or running an oracle-condition experiment in which the stage and reflection level are supplied by an expert or by a different model.
minor comments (6)
  1. [Introduction, first paragraph] There is a missing space in 'reassuringclients' in the first paragraph; it should read 'reassuring clients'.
  2. [§3.2.2, Eq. (2)] The subscript in 'lt t' is malformed and should be a single subscripted variable, e.g., l_t, to avoid confusion with the turn index t.
  3. [§4.2, Table 2] The sentence 'both evaluation approaches confirm INT's superiority in facilitating narrative transformation' is imprecise because INT's overall IM salience in Table 2 (35.064%) is lower than Doubao-1.5-pro's (42.110%); the claim should specify that INT shows the highest Level-2 salience in the automated setting.
  4. [Appendix D] The sentence describing the IM classification code book contains the typo 'IM typs' and should be 'IM types'.
  5. [Figure 4(a)] The caption of Fig. 4(a) does not explain which color corresponds to which variant or which bar group corresponds to which metric; the figure is difficult to read without that legend.
  6. [§4.4] The human evaluation comparing NTConv◦ and ESConv◦ reports win/loss/tie counts but no significance test; a simple sign test or bootstrap would strengthen the claim that NTConv◦ is preferred.

Circularity Check

1 steps flagged · score 6.0 of 10

IMA salience denominator includes therapist words, so the 'nearly doubling' of Level-2 IMs is partly a response-length artifact rather than a client-narrative effect.

  1. other [Section 3.3.3, Eq. (5); Section 4.3, Figure 4(a); Table 3]
    "Salience(I_i) = Σ_t WordCount(C_t ∩ I_i) / Σ_t WordCount(C_t ∪ T_t) ... Figure 4(a) shows INT produces focused responses (66.1 vs. 113.0 words), longer dialogues (57 vs. 42 turns), and richer user input (38.8 vs. 31.5 words) than directly role-playing with GPT-4o."

    The IMA salience metric is defined with the therapist's words in the denominator, so for a fixed amount of client IM content the reported percentage is inflated by short therapist replies and depressed by long ones. INT is explicitly instructed to keep replies brief ('Keep it short—no more than three sentences'), and the human evaluation shows INT averages 66.1 therapist words per turn versus 113.0 for GPT-4o role-play, with client turns of 38.8 versus 31.5 words. The paper's headline 'nearly doubling' Level-2 IMs (19.408% vs. 10.826% in Table 3) therefore partly reflects the denominator difference: converting to per-turn IM words gives roughly 20.4 vs. 15.6 (a 1.3x ratio), and a client-only denominator brings the comparison near parity (~52.5% vs. ~49.6%).

full rationale

The paper's derivation chain is otherwise not circular: INT's stage and reflection planning are not fitted to the evaluation target, the IM taxonomy is grounded in an external coding system (Gonçalves et al., 2011; Montesano et al., 2017), the human evaluation uses external expert annotators with reported inter-rater reliability, and no load-bearing self-citation chain is present. The one significant circularity-like reduction is the IMA salience metric of Eq. (5): because the denominator includes therapist words, the reported IMA superiority reduces in part by construction to the therapist response-length gap between INT and role-playing baselines. The paper neither reports absolute IM word counts nor client-only salience, so the 'nearly doubling' claim cannot be separated from this artifact. The remaining evidence (participant Likert ratings, downstream SFT comparison, and the acknowledged GPT-evaluation inflation) is independent, so the score is partial rather than total.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The ledger is light: no fitted constants drive the central claim and no novel physical entities are postulated. The main burdens are domain assumptions about the validity of the stage and reflection formalization, IM salience as an outcome proxy, and the use of GPT-4o for client simulation and evaluation.

free parameters (1)
  • k (number of retrieved exemplars) = 5
    Chosen by hand in Section 3.2.3; not fitted to data and likely minor, but it is a free design constant.
assumptions (5)
  • domain assumption White and Epston's four-stage narrative therapy process, with reflection levels per stage, can be faithfully encoded as a discrete planning state (S, L).
    The whole INT architecture assumes this formalization is clinically faithful; introduced in Section 3.2.1 and Figure 3.
  • domain assumption Salience of Innovative Moments, the word proportion of client speech classified as IM, is a valid measure of therapeutic progress.
    Used to compare systems and to claim nearly doubling of advanced IMs; based on Goncalves et al. (2011) but not independently validated here, Section 3.3.3.
  • domain assumption GPT-4o can simulate realistic help-seekers from ESConv profiles.
    Automated evaluation depends on this; the authors note simulated clients are too compliant, Section 4.1.3 and Section 4.2.
  • domain assumption GPT-4o can act as a strict counseling supervisor who scores therapeutic dimensions and IMs reliably.
    Table 2 uses this; the paper itself shows GPT evaluation overestimates scores relative to human evaluation, Section 4.1.3 and Section 4.2.
  • domain assumption Participant self-reports on four Likert dimensions capture therapeutic quality.
    The main human evidence is subjective ratings, with no clinical outcome or behavioral follow-up, Section 4.1.3 and Appendix C.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reframe Your Life Story: Interactive Narrative Therapist and Innovative Moment Assessment with Large Language Models." pith.science (2026). https://pith.science/paper/YZ34IYGQ

@misc{pith2026250720241,
  author       = {Pith},
  title        = {Pith review of: Reframe Your Life Story: Interactive Narrative Therapist and Innovative Moment Assessment with Large Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YZ34IYGQ}},
  note         = {Machine review of arXiv:2507.20241}
}
read the original abstract

Recent progress in large language models (LLMs) has opened new possibilities for mental health support, yet current approaches lack realism in simulating specialized psychotherapy and fail to capture therapeutic progression over time. Narrative therapy, which helps individuals transform problematic life stories into empowering alternatives, remains underutilized due to limited access and social stigma. We address these limitations through a comprehensive framework with two core components. First, INT (Interactive Narrative Therapist) simulates expert narrative therapists by planning therapeutic stages, guiding reflection levels, and generating contextually appropriate expert-like responses. Second, IMA (Innovative Moment Assessment) provides a therapy-centric evaluation method that quantifies effectiveness by tracking "Innovative Moments" (IMs), critical narrative shifts in client speech signaling therapy progress. Experimental results on 260 simulated clients and 230 human participants reveal that INT consistently outperforms standard LLMs in therapeutic quality and depth. We further demonstrate the effectiveness of INT in synthesizing high-quality support conversations to facilitate social applications.

Figures

Figures reproduced from arXiv: 2507.20241 by the authors.

Figure 1
Figure 1. An example dialogue shows how narrative therapy can help a single mother shift from viewing her disability and life as a despairing burden (problematic) to rediscovering purpose by recognizing her resilience in overcoming adversity (alternative). However, psychological distress often distorts these stories into problematic narratives, reinforc￾ing negative identities and vicious cycles (White and Epston, 1990). When… view at source ↗
Figure 2
Figure 2. Overview of our framework, comprising INT (left, green box) for narrative therapist simulation and IMA (right, blue box) for evaluating therapeutic progression. INT plans the therapeutic state (stage, reflection level) and generates responses aligned with expert style using retrieved exemplars. IMA classifies each client utterance into six IM types (e.g., Action I) across two levels and quantifies therapeutic progre… view at source ↗
Figure 3
Figure 3. Overview of reflection levels within each therapeutic stage of narrative therapy, including specific [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Statistics (a) and therapeutic state distribution (b) of interactive dialogues with INT variants. In (b), [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The IMs trajectory across dialogue turns (x [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

76 extracted references · 75 canonical work pages

  1. [1]

    Do not aim for long-term personality reconstruction; instead, help clients discover and apply their own resources

    work page

  2. [2]

    Change is driven by the client’s knowledge and lived experience—not by the therapist’s expertise

    work page

  3. [3]

    Emotions arise from narratives and evolve with them; emotions are not the only key to change

    work page

  4. [4]

    Focus on the connection between past, present, and future to help clients uncover new possibilities

    work page

  5. [5]

    Avoid jargon and remain sensitive to cultural backgrounds

    Speak the client’s language with empathy and respect. Avoid jargon and remain sensitive to cultural backgrounds

    work page

  6. [6]

    Significant others (e.g., family, friends) are crucial supporters in the client’s efforts to reshape their life story

    work page

  7. [7]

    discourses,

    Cultural and social “discourses,” especially from childhood, profoundly shape individual life experiences. Therapists should acknowledge and respect these influences and foster a more inclusive therapeutic space. Task Objective: You are to respond to the client by integrating the goals and strategies of the current therapeutic stage and the corre- spondin...

    work page

  8. [8]

    You adapt to the client’s language style, mirroring their way of speaking to ensure content is easily understandable, avoiding technical expressions

    work page

Show all 76 references
  1. [9]

    You avoid repetitive content in responses and refrain from asking rhetorical questions

  2. [10]

    Read the goals and strategies of the current therapeutic stage and reflection level

  3. [11]

    Study the example response to understand its tone and techniques

  4. [12]

    Rewrite the assistant’s initial reply using a similar tone and approach, including attentive listening, empathy, one guiding question, and gentle support

  5. [13]

    Keep it short—no more than three sentences, in a style resembling casual WeChat messages

  6. [14]

    Guidelines:

    Prompt the client to share more detailed experiences. Guidelines:

  7. [15]

    Help clients explore their local knowledge and resources

    Stay curious—never assume. Help clients explore their local knowledge and resources

  8. [16]

    Keep it informal, connected to real-life context, and avoid sounding like AI or academic writing

  9. [17]

    Action” and “Reflection

    (3)Supervised coding: Analysis of sample therapy transcripts with feedback from an experi- enced narrative therapist. (4)Calibration sessions: Group discussions where coders analyzed ambigu- ous cases to establish consistent interpretation. D.3 Reliability Assessment After com...

    2011

  10. [18]

    Match the client’s natural language style—clear, relatable, and never abstract or jargon-filled

  11. [19]

    Ask only one question at a time, progressing gently and naturally. 5. Speak like a caring friend, warmly and conversationally

  12. [20]

    Do not place emojis mid-sentence

    End every sentence with an emoji. Do not place emojis mid-sentence

  13. [21]

    decentered but influential

    When advice is needed, give it directly—don’t ask for permission to suggest. Table 14: System instruction for generating appropriate response given the dialogue history, retrieved expert examples, fixed reflection levelExploration of Identity Landscapewithin defined therapeuti...

  14. [22]

    You use a warm tone, employing positive responses and gentle prompts to encourage users to express more

  15. [23]

    If users are feeling down, you offer simple comfort and empathy without delving into professional therapy or diagnosis

  16. [24]

    During casual conversations, you can ask about interesting or shareable aspects of their daily lives to maintain a pleasant atmosphere

  17. [25]

    You maintain polite and friendly language, avoiding any offensive or inappropriate content

  18. [26]

    You closely follow White’s narrative therapy style: responses should be natural and coherent, matching the conversation context, using authentic colloquial English, avoiding official or rigid responses

  19. [27]

    Maintain curiosity about the client: respect their experiences and help them explore breakthrough points in addressing life challenges

  20. [28]

    You don’t assume users’ emotions as positive or negative, but wait for them to express themselves

  21. [30]

    Basic Assumptions of Narrative Therapy:

    Your responses should naturally transition from the user’s previous response, then directly engage in conversational narrative counseling based on their last answer. Basic Assumptions of Narrative Therapy:

  22. [31]

    No attempt at long-term personality reconstruction

  23. [32]

    No need for counselor’s professional knowledge; change requires the client’s own knowledge and resources

  24. [33]

    Emotions are not the only key to change; they arise within narratives and change as narratives evolve

  25. [34]

    Focus on connections between present, past, and future

  26. [35]

    Therapists learn to resonate with individual language to support them more effectively

    Therapists use the client’s language, avoiding complex theoretical terms, making support more aligned with the client’s cultural background. Therapists learn to resonate with individual language to support them more effectively

  27. [36]

    Family, peer groups, or other significant others are witnesses and supporters of the client’s life story

  28. [37]

    discourses

    Cultural and social “discourses” profoundly influence individual life experiences, especially childhood experiences. Therapists should focus on how these cultural discourses shape individuals’ worldviews and respect life stories within different cultural contexts, promoting mo...

  29. [38]

    Adjust the level of detail and openness in your response based on the set level of cooperation

  30. [39]

    Maintain natural, authentic, and coherent responses based on dialogue history

  31. [40]

    Express your emotions and inner world, including specific feelings, experiences, and thoughts

  32. [41]

    {last therapist response}

    You must have a reasonable psychological distress, with a low mood Avoid repeating content from dialogue history, keep it brief, no more than 30 words. User Prompt You are the client, please respond to the counselor based on the dialogue history and your level of cooperation. ...

  33. [42]

    Providing responses with warmth and calmness to emo- tionally intense disclosures when clients feel vulnerable

  34. [43]

    Effectively alleviating client anxiety and distress

  35. [44]

    Creating safety space to encourage openness and sharing

  36. [45]

    Validating feelings without minimizing clients’ struggles

  37. [46]

    Empowering The therapist’s capacity to help clients redefine and reframe their problems, particularly during the externalization phase

    Offering consistent emotional presence during the session. Empowering The therapist’s capacity to help clients redefine and reframe their problems, particularly during the externalization phase. In narrative therapy, externalization is a key strategy for helping clients separa...

  38. [47]

    Supporting clients externalize predominant problems and distinguish them from their personal identity

  39. [48]

    Helping clients examine problems from new perspectives

  40. [49]

    Promoting clients’ sense of self-efficacy

  41. [50]

    Guiding clients in setting achievable goals

  42. [51]

    Transformative The therapist’s ability to facilitate profound self-reflection, helping clients reconstruct their thinking or develop new cognition

    Providing constructive feedback and suggestions. Transformative The therapist’s ability to facilitate profound self-reflection, helping clients reconstruct their thinking or develop new cognition. In narrative therapy, re-authoring is a core phase where, through therapeutic in...

  43. [52]

    Facilitating clients’ cognitive restructuring and redefini- tion of past experiences through therapeutic guidance

  44. [53]

    Promoting multi-perspective self-reflection of clients

  45. [54]

    Guiding clients to discover new possibilities

  46. [55]

    Promoting behavioral change in clients

  47. [56]

    Assisting clients in constructing new narratives. Reconnecting The therapist’s ability to help clients re-examine their re- lationships with others, themselves, or their environment, with the goal of restoring and strengthening emotional con- nections. Re-membering is a crucia...

  48. [57]

    Helping clients re-examine relationships with significant others / things and environments

  49. [58]

    Promoting social connections for clients

  50. [59]

    Assisting clients in establishing new relationship patterns or restoring emotional harmony and balance

  51. [60]

    Humaneness The therapist agent’s ability to provide natural, fluid re- sponses that feel authentic, avoiding mechanical replies

    Promoting connection with important figures. Humaneness The therapist agent’s ability to provide natural, fluid re- sponses that feel authentic, avoiding mechanical replies. We expect therapist agent to mimic human interaction in their language, making the dialogue feel warm a...

  52. [61]

    Maintaining natural and fluid conversation

  53. [62]

    Responding in a natural and caring manner

  54. [63]

    Maintaining appropriate self-disclosure

  55. [64]

    Expressing empathy through human-like interaction

  56. [65]

    Table 19: Therapeutic dimension definitions and their evaluation criteria in narrative therapy

    Demonstrating professional humanistic care. Table 19: Therapeutic dimension definitions and their evaluation criteria in narrative therapy. System Instruction You are a rigorous and professional supervisor specializing in narrative therapy. Your role is to carefully evaluate t...

  57. [66]

    Scoring Guidelines (using 0.5-point intervals): • 1.0: Fails to meet any of the evaluation criteria • 1.5: Barely meets minimum requirements, significantly below standards • 2.0: Partially meets criteria but with notable deficiencies • 2.5: Meets some criteria but overall perf...

  58. [67]

    Evaluation Principles: • Strictly follow the specific criteria in the evaluation standards • Ensure score differentiation by appropriately using 0.5-point intervals • Provide clear justification for each scoring decision • Offer specific explanations that highlight both streng...

  59. [68]

    Please evaluate the following counseling dialogue based on the dimension of **{Dimension}** in narrative therapy practice

    Important Considerations: • Do not award high scores based solely on dialogue fluency • Strictly compare against the specific criteria in the evaluation standards • Provide thorough reasoning for each scoring point • Maintain objectivity and avoid subjective bias • Focus on th...

  60. [69]

    These moments represent the client’s beginning to question, challenge, or change their problem narrative

    IMCS Standard Definitions 1.1 Definition of Innovative Moments (IM) Innovative Moments (IM) are moments in therapeutic dialogue where the client’s narrative differs from the dominant problem narrative. These moments represent the client’s beginning to question, challenge, or c...

  61. [70]

    - When Action or Reflection (or both) overlap with Protest, code as Protest - Ensure IM content is complete and accurate

    Coding Rules and Standards 2.1 Basic Rules - When Action and Reflection coexist, code them separately. - When Action or Reflection (or both) overlap with Protest, code as Protest - Ensure IM content is complete and accurate. - When the client’s speech does not exhibit any IM c...

  62. [71]

    *client-generated*: Client actively generates without guidance

  63. [72]

    *therapist-prompted, client-elaborated*: Therapist questions or guides, client elaborates

  64. [73]

    *therapist-initiated, client-accepted*: Therapist provides explicit content or framework, client accepts

  65. [74]

    Narrative Dynamics Analysis Framework 3.1 Dominant Narrative Change - How has the original negative/restrictive narrative transformed? - What is the new narrative theme? - How has the narrative structure been reorganized? 3.2 Self-Identity Change - Has the client’s self-positi...

  66. [75]

    Output Format Please use the following Y AML format for output: ```YAML annotation: <IM type annotation> resource: <IM type resource> confidence: <confidence score> latent_narrative_dynamics_analysis: <analysis text> ```

  67. [76]

    Therapist: You mentioned feeling lonely recently, could you share your feelings about loneliness?

    Important Notes - Annotation must consider context and understand the overall dialogue flow. - Ensure IM content is complete, do not truncate sentences. - Narrative dynamics analysis should be specific and targeted, very concise and clear. - When uncertain, provide lower confi...

  68. [2016]

    Can you tell me what brought you here today?

    How not to evaluate your dialogue system: An empirical study of unsupervised evaluation met- rics for dialogue response generation. InProceedings of the 2016 Conference on Empirical Methods in Nat- ural Language Processing, pages 2122–2132. Siyang Liu, Chujie Zheng, Orianna De...

    2016 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.