Pith. sign in

REVIEW 4 major objections 5 minor 37 references

STAMPsy: Towards SpatioTemporal-Aware Mixed-Type Dialogues for Psychological Counseling

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that counseling chatbots improve when they choose the therapist's next helping skill and account for the client's time and place before replying.

desk verdict Useful dataset, but the effectiveness claim is entangled with how the data were generated; needs external validation. read the letter →

arxiv 2412.16674 v1 pith:7FOTX77H submitted 2024-12-21 cs.AI

Jieyi Wang , Yue Huang , Zeming Liu , Dexuan Xu , Chuan Wang , Xiaoming Shi , Ruiyuan Guan , Hongxing Wang
show 2 more authors
Weihua Yue Yu Huang
This is my paper · ORCID
classification cs.AI
keywords psychologicalcounselingdialoguemixed-typespatiotemporalawarenesshelpingskillsself-feedbacklargelanguagemodelsChinesedatasetretrieval-augmentedgeneration
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

The paper argues that online psychological-counseling chatbots should handle several dialogue types in one session and should know what they are doing and where and when the client is. To make this concrete, it introduces STAMPsy, a Chinese dataset of 5,006 counseling dialogues and 61,832 utterances in which every turn is annotated with the counselor's helping skill, the client's behavior, and a spatiotemporal stamp recording time, place, and weather. It then builds Self-STAMPsy, a framework that predicts the next helping skill before generating each reply, extracts the spatiotemporal state, retrieves relevant knowledge, and refines responses through self-reflection. The reported experiments on this benchmark show that the full model outperforms several fine-tuned LLM baselines on automatic metrics and on clinician-rated relevance, helpfulness, and empathy, and that removing the goal or spatiotemporal modules degrades performance. If these results transfer, goal clarification and world-awareness are practical design choices for AI counseling rather than optional extras.

What carries the argument

The load-bearing mechanism is the helping-skill sequence. Each counselor utterance is labeled with one of eight helping skills drawn from counseling psychology — open questions, restatements, feeling reflection, interpretation, information giving, direct guidance, immediacy, and self-disclosure — and the system predicts the next skill before writing the response, converting it into a natural-language instruction such as 'The therapist will then design a therapy.' The second mechanism is the spatiotemporal stamp: each dialogue is linked to a quadruple recording domain, slot, value, and time or place or weather stamp, and the model generates an emotional-state consequence from that stamp. These two signals are fed, together with adaptively retrieved knowledge-graph triples and an iterative self-feedback reflection, into a fine-tuned dialogue-generation LLM. The framework is called Self-STAMPsy, and the benchmark, STAMPsy, is the dataset plus the three sub-tasks used to test it.

What would settle it

A concrete test would be to label a small corpus of real anonymized online counseling sessions with the same helping-skills and spatiotemporal annotations and to rerun Self-STAMPsy against the strongest baselines under the paper's evaluation protocol; if the gap narrows or reverses, the central claim is not transferable.

Watch

Extended reading notes

Core claim

The central claim is that mixed-type counseling dialogues can be generated more faithfully when the model treats each counselor turn as an explicit helping-skill goal embedded in a spatiotemporal context. The authors report that a model that first predicts the helping skill and then uses the spatiotemporal stamp outperforms fine-tuned general-purpose and psychology-oriented baselines on automatic metrics and on clinician-rated relevance, informativeness, human-likeness, helpfulness, and empathy. Ablations show quality drops when the helping-skill prompt, the spatiotemporal stamp, or the self-feedback module is removed, with the empathy rating substantially lower when spatiotemporal information is omitted. The discovery is thus a mechanism-level result: clarifying dialogue goals in advance and stamping dialogues with time and place are effective interventions, not just dataset artifacts.

Load-bearing premise

The load-bearing premise is that expert-revised, LLM-generated dialogues from casebooks and public QA data resemble real counseling closely enough that quality gains measured on this benchmark transfer to actual online clients.

Editorial extensions

If this is right

  • Counseling systems can be designed as mixed-type conversations, so one model moves between diagnosis, information giving, empathetic support, recommendations, and question answering within a single session.
  • Clarifying the counselor's goal before each reply keeps conversations anchored to the client's emotional concerns, rather than drifting into generic educational or medical advice.
  • Spatiotemporal awareness lets a system tailor suggestions to the client's actual situation — for example, daylight exposure and a rich breakfast for early-morning awakening, versus pre-sleep relaxation for late-night insomnia.
  • The ablation results indicate that each of the four modules, helping-skill selection, spatiotemporal processing, adaptive retrieval, and iterative self-feedback, contributes to the reported quality; removing any one lowers performance.
  • A public Chinese benchmark now exists for comparing mixed-type psychological-counseling dialogue systems, with goal and spatiotemporal annotations as part of the standard evaluation.

Reading between the lines

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

  • Inference: the helping-skill prediction component may be useful on its own as a training or supervision aid for counselors, independent of the response generator, because it operates on transcripts and produces a checkable skill label.
  • Inference: the spatiotemporal quadruple mechanism could absorb other nonverbal context cues the paper lists but does not implement, such as paralanguage or body movement, if such inputs were available from a webcam or wearable device.
  • Inference: because the dataset is LLM-generated from casebooks and public question data and then expert-revised, a decisive test of transfer is whether the same gap over baselines appears on real, anonymized counseling transcripts; the paper itself notes that real consultation records are difficult to obtain.
  • Inference: the paper's own case study notes that the model sometimes asks several questions in one turn; goal clarification without a pacing constraint may add pressure, so a turn-level question-count control would be a natural extension.
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

4 major / 5 minor

Summary. The paper introduces STAMPsy, a Chinese dataset of 5,006 mixed-type psychological counseling dialogues with helping-skill annotations and spatiotemporal stamps, and proposes Self-STAMPsy, an iterative self-feedback framework that combines helping-skill selection (via BERT), spatiotemporal state processing, adaptive graph-based retrieval, and LoRA fine-tuning of Qwen2. The authors evaluate on STAMPsy and a small external GHSC transcript, reporting that Self-STAMPsy outperforms baselines on automatic metrics and manual ratings by psychotherapists, and ablation results indicate that removing the helping-skill prompt, spatiotemporal stamp, or self-feedback module degrades performance. The paper claims that clarifying dialogue goals in advance and utilizing spatiotemporal states are effective for psychological counseling dialogue generation.

Significance. If the dataset is trustworthy and the evaluation sound, the paper makes a useful contribution: it is the first mixed-type psychological counseling dataset with spatiotemporal awareness, it ships with a public repository, and the task formulation (goal clarification before counseling) is practical and under-explored. The expert revision and quality audit (0.91 quality score, Kappa 0.84) are commendable. However, the evaluation is currently entangled with the data-generation pipeline, and the quantitative evidence for the central claim is statistically thin. The core idea deserves to be published, but only after the evaluation is made independent and rigorous.

major comments (4)
  1. [Dataset Collection and Appendix F] The golden responses in STAMPsy are generated by an LLM using a multipart system prompt that embeds the helping-skill descriptions and the spatiotemporal-state tables (time of day, weather, location) that Self-STAMPsy later injects as e_gi and g_ST_i. The test split is drawn from the same corpus, so the ablations '-w/o Helping Skills' and '-w/o ST_i' in Table 6 primarily measure whether the model replicating the data-creation conditioning produces text closer to the gold, not whether those signals improve counseling quality. Expert revision checked internal consistency and correctness but did not independently validate the causal usefulness of the goal/stamp conditioning. The central claim that clarifying goals and using spatiotemporal states are effective therefore needs an evaluation not entangled with the data-generation pipeline, such as a human study on real or independently written counseling responses.
  2. [Experiments, Table 6] No significance tests or confidence intervals are reported for any automatic or manual metric. Several decisive comparisons are small: the GHSC gap between Self-STAMPsy (70.91) and GPT4 (69.09) is 1.8 points, and the manual Rel. and Help. scores are identical (1.85) or differ by 0.02–0.05 on a 0–2 scale. The manual-evaluation description states only that 'three out of four evaluators are consistent on the results of at least 910 conversations in the test set,' without the number of scored conversations, per-item variance, or inter-annotator agreement for the five indicators. Without this information, the reported improvements cannot be distinguished from noise.
  3. [Dataset Collection / Data Source] The authors state that they 'divide long cases into several meetings manually' but do not describe a case-level train/test split. If meetings from the same original case appear in both training and test sets, the model has already seen the client's profile and history, inflating absolute scores and reducing the validity of the benchmark. The paper should group all meetings derived from the same source case into the same split and report results on that split.
  4. [Evaluation / GHSC] The GHSC evaluation uses a single helper-client transcript from Hill's system to measure helping-skill understanding, and the annotation scheme is the same Hill taxonomy used to annotate STAMPsy. A single transcript, annotated under the same framework, is too small and too dependent on the STAMPsy protocol to support the general conclusion that 'current LLMs are unable to accurately clarify goals' (Table 6, GHSC column). Expanding GHSC to multiple transcripts with independent expert annotations would strengthen the claim.
minor comments (5)
  1. [Table 2] The 'Total 5330' appears to be the sum of the Direct Guidance subtypes (89+1128+2210+1777+137=5341), not the total of all goal types; the table caption should clarify what the numbers count and reconcile them with Table 3's 'Avg. # of goals per dialogue 16.51' (the five goal-type counts sum to 25,672, which would imply a different average).
  2. [Experiments] The sentence 'Following PsyCoun(Hu et al. 2024)' appears to refer to 'PsycoLLM' in the references; check the citation.
  3. [Methodology] Figure 4 is not referenced in the text; add a reference such as '(see Figure 4)'.
  4. [Case Study and Table 6] There are typos: 'SpatioTmeporal' in the Table 6 caption, 'environemnt' in the case study, and a duplicated sentence in the Case Study paragraph about MindChat and EmoLLM.
  5. [Appendix I] The manual evaluation guideline says 'score 0 (bad): not fluent and the answer is not in line with the question all'; fix the grammar. Also, the data quality audit reports a Kappa for the 1/0 quality evaluation but not for the fine-grained helping-skill and goal annotations; consider reporting that agreement.

Circularity Check

1 steps flagged · score 6.0 of 10

STAMPsy's gold responses were generated with the same helping-skill and spatiotemporal conditioning that Self-STAMPsy injects, so the Table 6 ablations partly reward pipeline replication; the spatiotemporal-effectiveness claim lacks external validation.

  1. other [Dataset Collection (Appendix C); Methodology (SpatioTemporal Stamp Processing and Evaluation)]
    "Based on reference knowledge quadruples, we conclude a multipart system prompt shaped by instructions based on the CCM and expand single-turn text to multi-turn dialogues. ... we have invited professional psychological counselors to summarize the potential effects of this information on emotional states and integrate these insights into the prompts. This allows for the creation of spatiotemporal stamps gST i in conjunction with the spatiotemporal state STi. ..."

    The causal chain is closed by the data-generation procedure. STAMPsy gold responses are not recordings of real counseling; they are LLM expansions produced under a multipart prompt that explicitly instructs helper personality, lists Hill helping-skill definitions, and contains expert-built tables tying time/weather/season/location to counseling responses and emotional states. Self-STAMPsy then conditions generation on predicted helping-skill prompts egi and spatiotemporal stamps gST_i derived from the same annotation scheme, and is evaluated against those gold responses on BLEU/ROUGE/BertSim and manual metrics (Table 6). The '-w/o Helping Skills' and '-w/o ST_i' ablations therefore reward models that re-instantiate the exact conditioning used to author the gold text.

full rationale

The paper is otherwise a straightforward dataset-and-baseline paper. The GHSC transcript provides independent evidence for helping-skill classification (Self-STAMPsy 70.91 vs GPT4 69.09), and the BERT classifier results are standard supervised learning. The circularity is limited to the causal-effectiveness claim drawn from within-corpus ablations; it does not invalidate the dataset as a resource, but it does mean the abstract's 'results indicate...' should be read as 'the model reproduces the generation-time conditioning.' Also, no case-level train/test split is described, so leakage from the same source case across meetings may inflate absolute numbers; this is a data-split concern, not a circularity per se. No load-bearing self-citations were found; citations to prior work are contextual. The circularity is in the evaluation design, not in citation practice.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The paper's central claims rest on several domain assumptions about counseling theory and spatiotemporal effects, plus a small number of undisclosed training hyperparameters. The main invented conceptual entity is the spatiotemporal stamp, an annotation object rather than a physical postulate. No free parameters are fitted to data in a way that would represent hidden degrees of freedom in a derivation.

free parameters (3)
  • BERT classifier hyperparameters (batch size 128, learning rate 0.001) = 128, 0.001
    Chosen by hand for the helping-skills classification module; not the focus of the paper, but affects goal-prediction quality and thus downstream generation.
  • Number of self-feedback iterations = not reported
    The iterative self-feedback mechanism's stopping condition is not specified, and response quality likely depends on this choice.
  • LoRA configuration (rank and alpha) = not reported
    The rank and alpha used for fine-tuning are not disclosed, making exact reproduction of the results in Table 6 incomplete.
assumptions (4)
  • domain assumption Hill's Helping Skills taxonomy is a valid and sufficient annotation schema for counselor goals in psychological counseling.
    Used to label every dialogue turn and to build the GHSC evaluation; if the taxonomy is not appropriate for online counseling, the goal-clarification claim is weakened.
  • ad hoc to paper LLM-generated dialogues, revised by experts, faithfully represent real counseling interactions.
    Privacy prevents real records; the dataset is generated from case books and PsyQA. This is the weakest premise underlying the dataset's external validity.
  • domain assumption Spatiotemporal state (time, location, weather) systematically influences appropriate counseling responses.
    Operationalized in Tables 8-11 and used to create spatiotemporal stamps; the paper cites supporting literature but assumes the derived mappings are correct.
  • domain assumption The 0-2 manual scoring rubric measures counseling quality.
    Used for the main evaluation claims; no relation to clinical outcomes is established, and inter-rater reliability for the generation evaluation is not reported.
invented entities (1)
  • Spatiotemporal stamp (g_ST_i)
    purpose: A prompt-level summary of how time, location, and weather affect the client's emotional state, used to condition response generation.
    It is a paper-defined annotation construct with no falsifiable handle outside the STAMPsy dataset; its validity depends on expert-authored heuristics and the cited psychological literature.

how reviews work

0 comments
Cite this review

Pith. "Pith review of STAMPsy: Towards SpatioTemporal-Aware Mixed-Type Dialogues for Psychological Counseling." pith.science (2026). https://pith.science/paper/7FOTX77H

@misc{pith2026241216674,
  author       = {Pith},
  title        = {Pith review of: STAMPsy: Towards SpatioTemporal-Aware Mixed-Type Dialogues for Psychological Counseling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7FOTX77H}},
  note         = {Machine review of arXiv:2412.16674}
}
read the original abstract

Online psychological counseling dialogue systems are trending, offering a convenient and accessible alternative to traditional in-person therapy. However, existing psychological counseling dialogue systems mainly focus on basic empathetic dialogue or QA with minimal professional knowledge and without goal guidance. In many real-world counseling scenarios, clients often seek multi-type help, such as diagnosis, consultation, therapy, console, and common questions, but existing dialogue systems struggle to combine different dialogue types naturally. In this paper, we identify this challenge as how to construct mixed-type dialogue systems for psychological counseling that enable clients to clarify their goals before proceeding with counseling. To mitigate the challenge, we collect a mixed-type counseling dialogues corpus termed STAMPsy, covering five dialogue types, task-oriented dialogue for diagnosis, knowledge-grounded dialogue, conversational recommendation, empathetic dialogue, and question answering, over 5,000 conversations. Moreover, spatiotemporal-aware knowledge enables systems to have world awareness and has been proven to affect one's mental health. Therefore, we link dialogues in STAMPsy to spatiotemporal state and propose a spatiotemporal-aware mixed-type psychological counseling dataset. Additionally, we build baselines on STAMPsy and develop an iterative self-feedback psychological dialogue generation framework, named Self-STAMPsy. Results indicate that clarifying dialogue goals in advance and utilizing spatiotemporal states are effective.

Figures

Figures reproduced from arXiv: 2412.16674 by the authors.

Figure 1
Figure 1. An example of STAMPsy with spatiotemporal state and reference knowledge. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The collection workflow of STAMPsy. We extract reference knowledge from context under 9-Box CCM and then [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Sequence visualization of the common dialogue [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: The framework of the proposed Self-STAMPsy. A detailed prompt template is open-sourced. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: A case of generated answers and the golden answer. It can be observed that without the guidance of counseling [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: A case of generated answers and the golden answer, focusing on the spatiotemporal state. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: The framework of Meichenbaum’s 9-Box Generic [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 32 canonical work pages

  1. [1]

    Explicit Content: What did the psychological counsel- ing client talk about?

    work page

  2. [2]

    Implicit Content: Is there any underlying meaning to what the psychological counseling client talked about?

    work page

  3. [3]

    Defense and Barriers to Change: How does the psycho- logical counseling client avoid anxiety?

    work page

  4. [4]

    Psychological Counseling Client’s Distortion: In what ways does the psychological counseling client’s reaction to you mirror their reactions to significant others in their life?

    work page

  5. [5]

    ArXiv, abs/2211.05100

    BLOOM: A 176B-Parameter Open-Access Multilin- gual Language Model. ArXiv, abs/2211.05100. Sherry, C.; Paula, N.; and Cynthia, O. 2009. Interviewing and Change Strategies for Helpers: Fundamental Skills and Cognitive-Behavioral Interventions. Brooks/Cole, 6th edi- tion. Instructor’s Edition. Shi, X.; Liu, Z.; Wang, C.; Leng, H.; Xue, K.; Zhang, X.; and Zha...

    arXiv 2009

  6. [6]

    Recommendation: Treatment

    Personal Assessment: How do you evaluate your re- sponse? If possible, what different responses would you make? Why? Manual Annotation Guidiance We provide the complete content of the conversation, and the annotator needs to infer the time and location of the conver- sation based on its content and annotate it accordingly. Then, summarize the chief compla...

    work page 1992

  7. [7]

    [Online; accessed: 2024-08-12]

    Web page. [Online; accessed: 2024-08-12]. Xin Yan, D. X. 2023. MindChat: Psychological Large Lan- guage Model. https://github.com/X-D-Lab/MindChat. Yang, A.; Xiao, B.; Wang, B.; Zhang, B.; et al

    work page 2024

  8. [8]

    arXiv:2309.10305

    Baichuan 2: Open Large-scale Language Models. arXiv:2309.10305. Yang, S.; Zhang, R.; and Erfani, S. 2020. GraphDialog: In- tegrating Graph Knowledge into End-to-End Task-Oriented Dialogue Systems. In Proceedings of the 2020 Confer- ence on Empirical Methods in Natural Language Processing (EMNLP). Yao, B.; Shi, C.; Zou, L.; Dai, L.; Wu, M.; Chen, L.; Wang,...

    arXiv 2020

Show all 37 references
  1. [9]

    Computer Generated Report Based on the Case Conceptualization Model

    AugESC: Dialogue Augmentation with Large Lan- guage Models for Emotional Support Conversation. InFind- ings of ACL. Zhou, H.; Xu, X.; Wu, W.; Niu, Z.; Wu, H.; Bao, S.; Wang, F.; and Wang, H. 2022. Link the world: Improv- ing open-domain conversation with dynamic spatiotempo- r...

    2022 arXiv

  2. [10]

    (Self-disclosure) Do you feel the same way?(Questions) Client: Um, I don’t know

    Helper: When I left home, my parents were very sad, and I felt bad and guilty. (Self-disclosure) Do you feel the same way?(Questions) Client: Um, I don’t know. It’s hard to put everything into words

  3. [11]

    But I’m also angry because they don’t want me to grow up

    Helper: How do you feel about this situation?(Questions) Client: I feel guilty about wanting to leave them. But I’m also angry because they don’t want me to grow up. I know they have problems, but they should solve them themselves. What do you think I should do?

  4. [12]

    (Direct Guidance) You also need to talk to your parents and tell them how you feel.(Direct Guidance) Client: Well, I will implement it

    Helper: You should move out. (Direct Guidance) You also need to talk to your parents and tell them how you feel.(Direct Guidance) Client: Well, I will implement it. What should I do if I want to live in the dormitory?

  5. [13]

    Do you really think I should move out?

    Helper: The school’s housing office provides all the rel- evant information, (Information Giving) it’s on the other side of the school.(Information Giving) Client: I think I should give them a call. Do you really think I should move out?

  6. [14]

    Countertransference: In what ways have your emotions, attitudes, and behavioral responses been stimulated by your interactions with the psychological counseling client?

  7. [15]

    You might be right

    Helper: I’m a bit surprised that you want me to tell you what to do, (Immediacy) I wonder if you also let your parents tell you what to do, and then you get angry when they do?(Interpretations) Client: I’ve never thought about it that way. You might be right. I do become passi...

  8. [16]

    (Others) My name is Judy

    Helper: Thank you for coming today. (Others) My name is Judy. (Information Giving) I am studying the art of helping. (Information Giving) We can talk for 20 minutes today. (Information Giving) You can talk about anything you want to talk about.(Direct Guidance) Client: Lately,...

  9. [17]

    (Questions) By the way, what is your ma- jor?(Questions) Client: I haven’t decided on a major yet because I don’t know what I’m interested in

    Helper: Give me an example of the last time you didn’t go to school. (Questions) By the way, what is your ma- jor?(Questions) Client: I haven’t decided on a major yet because I don’t know what I’m interested in

  10. [18]

    (Restatments) Do you live on campus?(Questions) Client: I live at home and I feel a lot of pressure

    Helper: So you haven’t decided yet. (Restatments) Do you live on campus?(Questions) Client: I live at home and I feel a lot of pressure. I want to live in the dormitory, but my parents won’t pay for it, and I don’t have the money myself. I mean, my parents live very close to t...

  11. [19]

    I really hate it

    Helper: It sounds like your parents are forcing you to live at home.(Restatments) Client: Yes, that’s exactly how it is. I really hate it. I think if I lived in the dormitory, I would feel much freer. I feel very restricted at home, like they’re watching my every move, and I c...

  12. [20]

    Helper: You feel suffocated. (Reflection of feeling) It sounds like you’re very uncomfortable because your par- ents are too restrictive.(Reflection of feeling) Client: Yes, but I don’t know how to deal with it. They do provide me with a place to live and help me outside of sc...

  13. [21]

    (Information Giv- ing) I think you might be feeling a bit uneasy

    Helper: You just became very restless, (Information Giv- ing) and your voice also became soft. (Information Giv- ing) I think you might be feeling a bit uneasy. (Re- flection of feeling) Did I correctly express your feel- ings?(Questions) Client: Well, I feel very bad, like I’...

  14. [22]

    When I brought it up, they were all very sad, especially my mom

    Helper: How do you feel about that?(Questions) Client: Last night when they said they really didn’t want me to leave, I became very angry. When I brought it up, they were all very sad, especially my mom

  15. [23]

    (Interpretations) Perhaps they are not ready for you to leave home

    Helper: I wonder if neither you nor your parents can handle separation very well because your roles have changed, and you’ve grown up. (Interpretations) Perhaps they are not ready for you to leave home. Because when facing an empty house, they will be very anxious. (Inter- pre...

  16. [24]

    Helper: On the one hand, it’s hard for you to leave them; (Restatments) on the other hand, you really want to leave and live your own life.(Challenge) Client: Well, I want to move out, but I don’t want to hurt them

  17. [29]

    Helper: I know you want me to tell you what to do, (Im- mediacy) but I feel a bit anxious about giving you direct advice because I don’t know your situation very well, (Immediacy) you have to decide for yourself whether to move out or not.(Challenge) Client: I’m afraid of maki...

  18. [31]

    (Self-disclosure) For me, talking with them was very important in order to maintain our relationship

    Helper: I can tell you, that when I had problems with my parents, I had a serious talk with them and then moved out. (Self-disclosure) For me, talking with them was very important in order to maintain our relationship. (Self- disclosure) When I first sat down to talk with them...

  19. [32]

    Helper: How do you think you will act?(Questions) Client: Maybe I will talk to my parents

  20. [33]

    (Direct Guidance) Research shows that practicing in a helping conversation makes it easier for you to do it outside

    Helper: Let’s try it now. (Direct Guidance) Research shows that practicing in a helping conversation makes it easier for you to do it outside. (Information Giving) What I want you to do is, assume your parents are here, and you tell them you want to move out.( Direct Guidance)...

  21. [34]

    Helper: That’s a good start, (Others) but you hesitated a few times, (Information Giving) and your voice was too soft, (Information Giving) try to speak up and clearly tell them what you want.(Direct Guidance) Client: Mom and Dad, I’ve decided to move out

  22. [35]

    (Others) Your voice is loud and clear, (Information Giving) and you clearly ex- pressed what you want

    Helper: That sounds very good. (Others) Your voice is loud and clear, (Information Giving) and you clearly ex- pressed what you want. (Information Giving) Try to do that when you talk to them. (Direct Guidance) I want to tell you that I really enjoyed working with you because ...

  23. [36]

    (Others) Goodbye, (Others) I hope you enjoy the rest of your day

    Helper: That’s great. (Others) Goodbye, (Others) I hope you enjoy the rest of your day. (Others) Client: You too. Goodbye. Guideline for Human Evaluation To better evaluate the quality of the model’s responses, we conducted evaluations in five aspects, as illustrated in Table

  24. [37]

    In every data labeling, we conduct training in advance so that the labelers can understand accurately and evaluate cor- rectly

    Three out of four evaluators are consistent on the results of at least 910 conversations in the test set. In every data labeling, we conduct training in advance so that the labelers can understand accurately and evaluate cor- rectly. Finally, we will manually extract a portion...

  25. [2007]

    Routledge, 1st edi- tion

    Practicing Counseling and Psychotherapy: Insights from Trainees, Supervisors and Clients . Routledge, 1st edi- tion. Lee, S.; Seo, Y .; and Kim, D. 2007. Validation of the coun- selor activity self-efficacy scales. Korean J Couns Psy- chother, 19: 655–73. Lent, R.; Hill, C.; a...

    2007 arXiv

  26. [2020]

    In Proceedings of the 37th International Confer- ence on Machine Learning , ICML’20

    REALM: retrieval-augmented language model pre- training. In Proceedings of the 37th International Confer- ence on Machine Learning , ICML’20. JMLR.org. Hill, C. E. 1986. An overview of the Hill counselor and client verbal response modes category systems. In Green- berg, L. S.;...

    1986 arXiv

  27. [2022]

    In Proceedings of the 60th Annual Meeting of the Association for Computa- tional Linguistics (V olume 1: Long Papers), 1024–1034

    Where to Go for the Holidays: Towards Mixed-Type Dialogs for Clarification of User Goals. In Proceedings of the 60th Annual Meeting of the Association for Computa- tional Linguistics (V olume 1: Long Papers), 1024–1034. Madotto, A.; Lin, Z.; Bang, Y .; and Fung, P. 2020. The A...

    2020 arXiv

  28. [2023]

    ArXiv, abs/2310.11511

    Self-RAG: Learning to Retrieve, Generate, and Cri- tique through Self-Reflection. ArXiv, abs/2310.11511. Bai, J.; Bai, S.; Chu, Y .; Cui, Z.; Dang, K.; Deng, X.; Fan, Y .; Ge, W.; Han, Y .; Huang, F.; et al. 2023. Qwen technical report. arXiv preprint arXiv:2309.16609. Bickmor...

    2023 arXiv

  29. [2024]

    arXiv preprint arXiv:2404.11978

    EVIT: Event-Oriented Instruction Tuning for Event Reasoning. arXiv preprint arXiv:2404.11978. Tinsley, H. E.; and Weiss, D. J. 1975. Interrater reliability and agreement of subjective judgments. Journal of Counsel- ing Psychology, 22(4): 358–376. Touvron, H.; Lavril, T.; Izaca...

    1975 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.