SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2305.13112 · EMNLP · 2023

Rethinking the Evaluation for Conversational Recommendation in the Era of Large Language Models

Ji-Rong Wen, Wayne Zhao, Xinyu Tang, Xiaolei Wang, Jingyuan Wang

arXiv · PDF · Open in the Atlas

Code that ran

5 of 17 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 1 failure of our instrument; 12 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2305.13112.json · badge: /api/ran/2305.13112.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
rucaibox/ievalm-crs canonical (classifier label)
classifier basis: census
3 ran · 8 no recorded run
RUCAIBox/iEvaLM-CRS community
Papers with Code archive
2 ran · 4 no recorded run
FunctionStatusWhere it lives
annotate Ran
ran; our contract draft was wrong, not the code
RUCAIBox/iEvaLM-CRS/src/model/CHATGPT.py
code served (permissive licence) · get_code("1052ac20f3dfe2dc")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import typing
from unittest.mock import Mock, MagicMock, patch
import openai

# Mock the openai module and its components
openai.Embedding = Mock()
openai.error = Mock()
openai.error.Timeout = Exception
openai.error.InvalidRequestError = Exception
openai.error.AuthenticationError = Exception

# Create a mock response object
mock_response = {
    'object': 'list',
    'data': [
        {'object': 'embedding', 'embedding': [0.1, 0.2, 0.3], 'index': 0}
    ],
    'model': 'text-embedding-ada-002',
    'usage': {'prompt_tokens': 5, 'total_tokens': 5}
}

# Mock the create method to return the response
openai.Embedding.create = Mock(return_value=mock_response)

ARGS = {
    'conv_str': 'Hello, this is a test string for embedding annotation.'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
loguru, openai, tenacity — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
annotate_completion Ran
ran on a synthesized input
rucaibox/ievalm-crs/script/ask.py
code served (permissive licence) · get_code("bcfca1295d83a6e4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import typing
from unittest.mock import Mock, patch, MagicMock
import openai

# Mock the openai module to avoid actual API calls
openai.Completion = Mock()
openai.error = Mock()
openai.error.Timeout = Exception
openai.error.InvalidRequestError = Exception
openai.error.AuthenticationError = Exception

# Create a mock response that the function expects
mock_response = {
    'choices': [
        {
            'text': 'This is a test completion response.'
        }
    ]
}

# Patch openai.Completion.create to return our mock response
openai.Completion.create = Mock(return_value=mock_response)

# Define ARGS with all parameters for annotate_completion
ARGS = {
    'prompt': 'What is the meaning of life?',
    'logit_bias': {'100': 0.5, '200': -0.5}
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
openai, tenacity — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_instruction Ran
ran on a synthesized input
rucaibox/ievalm-crs/script/chat.py
code served (permissive licence) · get_code("8cfed84e6fae7269")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

ARGS = {
    'dataset': 'redial_conversation_v1'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
my_wait_exponential Ran
ran on a synthesized input · constructed an object; no result was computed
RUCAIBox/iEvaLM-CRS/src/model/CHATGPT.py
code served (permissive licence) · get_code("2419551feec87f10")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, my_wait_exponential: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import typing
from tenacity import _utils, Retrying, retry_if_not_exception_type
from tenacity.wait import wait_base
from unittest.mock import Mock

class my_wait_exponential(wait_base):

    def __init__(self, multiplier: typing.Union[int, float]=1, max: _utils.time_unit_type=_utils.MAX_WAIT, exp_base: typing.Union[int, float]=2, min: _utils.time_unit_type=0) -> None:
        self.multiplier = multiplier
        self.min = _utils.to_seconds(min)
        self.max = _utils.to_seconds(max)
        self.exp_base = exp_base

    def __call__(self, retry_state: 'RetryCallState') -> float:
        if retry_state.outcome == None:
            return 0
        try:
            exp = self.exp_base ** (retry_state.attempt_number - 1)
            result = self.multiplier * exp
        except OverflowError:
            return self.max
        return max(max(0, self.min), min(result, self.max))


# Create a mock retry_state for testing
mock_retry_state = Mock()
mock_retry_state.outcome = None
mock_retry_state.attempt_number = 3

ARGS = {
    'multiplier': 1.5,
    'max': 60,
    'exp_base': 2,
    'min': 1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
openai, tenacity — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type my_wait_exponential — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a my_wait_exponential output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
sample_data Ran
ran on a synthesized input
rucaibox/ievalm-crs/src/model/utils.py
code served (permissive licence) · get_code("82d4d951d061c46e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import random
import torch

torch.manual_seed(0)

# Create a sample data_list with diverse elements
data_list = [
    {"id": 0, "text": "sample 0"},
    {"id": 1, "text": "sample 1"},
    {"id": 2, "text": "sample 2"},
    {"id": 3, "text": "sample 3"},
    {"id": 4, "text": "sample 4"},
    {"id": 5, "text": "sample 5"},
    {"id": 6, "text": "sample 6"},
    {"id": 7, "text": "sample 7"},
    {"id": 8, "text": "sample 8"},
    {"id": 9, "text": "sample 9"},
] * 50  # Create a list with 500 elements to allow various shot values

ARGS = {
    "data_list": data_list,
    "shot": 1,
    "debug": False,
    "number_for_debug": 320,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
CHATGPT Not yet run
unverified (harvested, no recorded run)
RUCAIBox/iEvaLM-CRS/src/model/CHATGPT.py
code served (permissive licence) · get_code("7fab3666beecf259")
How it ran

no invocation recorded.

annotate Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/script/cache_item.py
code served (permissive licence) · get_code("4b4ed6d15480f5a5")
How it ran

no invocation recorded.

annotate Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/src/model/CHATGPT.py
code served (permissive licence) · get_code("285d00ffcd75ae67")
How it ran

no invocation recorded.

annotate_chat Not yet run
unverified (harvested, no recorded run)
RUCAIBox/iEvaLM-CRS/src/model/CHATGPT.py
code served (permissive licence) · get_code("e92d3fa7cbc2a58c")
How it ran

no invocation recorded.

annotate_chat Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/src/model/CHATGPT.py
code served (permissive licence) · get_code("a3d50124d306d006")
How it ran

no invocation recorded.

get_instruction Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/script/ask.py
code served (permissive licence) · get_code("c0900257616e5ccb")
How it ran

no invocation recorded.

get_model_args Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/script/ask.py
code served (permissive licence) · get_code("ed9662882a67bfc4")
How it ran

no invocation recorded.

get_model_args Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/script/chat.py
code served (permissive licence) · get_code("582ed53a923214b4")
How it ran

no invocation recorded.

load_jsonl_data Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/src/model/utils.py
code served (permissive licence) · get_code("4fe63c6cb46bf907")
How it ran

no invocation recorded.

my_before_sleep Not yet run
unverified (harvested, no recorded run)
RUCAIBox/iEvaLM-CRS/src/model/CHATGPT.py
code served (permissive licence) · get_code("fcd2cd328d39b785")
How it ran

no invocation recorded.

my_stop_after_attempt Not yet run
unverified (harvested, no recorded run)
RUCAIBox/iEvaLM-CRS/src/model/CHATGPT.py
code served (permissive licence) · get_code("29efbd3fe0422744")
How it ran

no invocation recorded.

simple_collate Not yet run
unverified (harvested, no recorded run)
rucaibox/ievalm-crs/src/model/utils.py
code served (permissive licence) · get_code("f1a7898875d9e106")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 15 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The recent success of large language models (LLMs) has shown great potential to develop more powerful conversational recommender systems (CRSs), which rely on natural language conversations to satisfy user needs. In this paper, we embark on an investigation into the utilization of ChatGPT for CRSs, revealing the inadequacy of the existing evaluation protocol. It might overemphasize the matching with ground-truth items annotated by humans while neglecting the interactive nature of CRSs. To overcome the limitation, we further propose an interactive Evaluation approach based on LLMs, named iEvaLM, which harnesses LLM-based user simulators. Our evaluation approach can simulate various system-user interaction scenarios. Through the experiments on two public CRS datasets, we demonstrate notable improvements compared to the prevailing evaluation protocol. Furthermore, we emphasize the evaluation of explainability, and Chat-GPT showcases persuasive explanation generation for its recommendations. Our study contributes to a deeper comprehension of the untapped potential of LLMs for CRSs and provides a more flexible and realistic evaluation approach for future research about LLMbased CRSs. The code is available at https: //github.com/RUCAIBox/iEvaLM-CRS.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2305.13112")
get_code_for_paper("2305.13112")
have("2305.13112")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2305.13112.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2305.13112.svg)](https://syntology.ai/paper/2305.13112)

Connect an agent — have() is free.