uid stringlengths 3 5 | uid_n int64 1 3 | abstract stringclasses 90
values | treatment stringclasses 4
values | original_abstract stringclasses 90
values | final_abstract stringlengths 458 2.92k ⌀ | use_genai_writing stringclasses 4
values | belief_use_genai_writing stringclasses 4
values | quality_genai_generating float64 0 5 ⌀ | quality_genai_editing float64 1 5 ⌀ | gender stringclasses 2
values | education stringclasses 4
values | profession stringlengths 2 68 ⌀ | age stringclasses 5
values | would_you_use_gpt? stringclasses 3
values | would_people_use_gpt? stringclasses 3
values | can_distinguish? stringclasses 4
values | distinguish_explanation stringlengths 2 752 ⌀ | frequency_of_ai_use int64 0 3 | trust_ai int64 0 6 | frequency_of_ai_in_writing int64 0 3 | edits int64 0 4.54k | confidence_edited int64 0 50 | confidence_original int64 0 50 | kinds stringclasses 31
values | quality int64 0 4 | readability_reported int64 2 100 | readability_original float64 -14.88 121 | readability_final float64 -79.55 121 | reviewer_decision_1 float64 0 100 ⌀ | reviewer_decision_2 float64 0 100 ⌀ | reviewer_decision_3 float64 0 100 ⌀ | reviewer_confidence_1 float64 0 100 ⌀ | reviewer_confidence_2 float64 0 100 ⌀ | reviewer_confidence_3 float64 10 100 ⌀ | reason_1 stringlengths 4 695 ⌀ | reason_2 stringlengths 4 933 ⌀ | reason_3 stringlengths 4 718 ⌀ | final_decision stringclasses 2
values | abstract_edit_timeseries_analysis listlengths 1 4.54k |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
p-1 | 1 | Deep neural networks are known to be highly vulnerable to adversarial attacks, where small, deliberate perturbations in input data can cause significant errors in classification. A key contributor to this vulnerability is the non-Lipschitz behavior of neural networks, which enables small input changes to yield large fe... | AI-noInfo | Deep neural networks are known to be highly vulnerable to adversarial attacks, where small, deliberate perturbations in input data can cause significant errors in classification. A key contributor to this vulnerability is the non-Lipschitz behavior of neural networks, which enables small input changes to yield large fe... | Deep neural networks are known to be highly vulnerable to adversarial attacks, where small, deliberate perturbations in input data can cause significant errors in classification. A key contributor to this vulnerability is the non-Lipschitz behavior of neural networks, which enables small input changes to yield large fe... | generate_content | 50-75% | 2 | null | Male | A bachelor's degree | Data scientist | 26-35 | Yes | Yes | Sometimes | AI-generated writings frequently have a more uniform, organized style with less emotional depth and complexity, whereas human-written texts generally have more diverse sentence patterns, personal tone, and nuanced emotions. | 2 | 2 | 2 | 20 | 19 | 41 | Correction | 4 | 90 | 1.683106 | 2.627918 | 25 | 75 | 75 | null | null | null | null | null | null | Accept | [
{
"time": "11:26:52",
"step": 1,
"revision_id": "A0",
"timestamp": "2025-04-14T11:26:52",
"full_text": " \nDeep neural networks are known to be highly vulnerable to adversarial attacks, where small, deliberate perturbations in input data can cause significant errors in classification. A key cont... |
p-1 | 2 | LLM-based code agents have advanced AI-assisted software development by combining large language models with external tools for dynamic code execution and system interaction. However, these agents pose significant safety risks, as they may generate or execute harmful code with severe security implications. Existing res... | AI-noInfo | LLM-based code agents have advanced AI-assisted software development by combining large language models with external tools for dynamic code execution and system interaction. However, these agents pose significant safety risks, as they may generate or execute harmful code with severe security implications. Existing res... | LLM-based code agents have advanced AI-assisted software development by combining large language models with external tools for dynamic code execution and system interaction. However, these agents pose significant safety risks, as they may generate or execute harmful code with severe security implications. Existing res... | generate_content | 50-75% | 2 | null | Male | A bachelor's degree | Data scientist | 26-35 | Yes | Yes | Sometimes | AI-generated writings frequently have a more uniform, organized style with less emotional depth and complexity, whereas human-written texts generally have more diverse sentence patterns, personal tone, and nuanced emotions. | 2 | 2 | 2 | 1 | 14 | 45 | Correction | 4 | 90 | 18.38355 | 18.96879 | 75 | 75 | 75 | null | null | null | null | null | null | Accept | [
{
"time": "11:29:05",
"step": 1,
"revision_id": "A0",
"timestamp": "2025-04-14T11:29:05",
"full_text": " \nLLM-based code agents have advanced AI-assisted software development by combining large language models with external tools for dynamic code execution and system interaction. However, these... |
p-1 | 3 | The rapid advancement of diffusion models has significantly improved content generation across various modalities, yet video generation still lags in visual quality due to noisy pre-training data and the lack of alignment with human preferences. To address these challenges, we introduce InstructVideo, a novel framework... | AI-noInfo | The rapid advancement of diffusion models has significantly improved content generation across various modalities, yet video generation still lags in visual quality due to noisy pre-training data and the lack of alignment with human preferences. To address these challenges, we introduce InstructVideo, a novel framework... | The rapid advancement of diffusion models has significantly improved content generation across various modalities, yet video generation still lags in visual quality due to noisy pre-training data and the lack of alignment with human preferences. To address these challenges, we introduce InstructVideo, a novel framework... | generate_content | 50-75% | 2 | null | Male | A bachelor's degree | Data scientist | 26-35 | Yes | Yes | Sometimes | AI-generated writings frequently have a more uniform, organized style with less emotional depth and complexity, whereas human-written texts generally have more diverse sentence patterns, personal tone, and nuanced emotions. | 2 | 2 | 2 | 16 | 21 | 35 | Correction | 4 | 90 | -6.48756 | -6.48756 | 75 | 75 | 75 | null | null | null | null | null | null | Accept | [
{
"time": "11:29:21",
"step": 1,
"revision_id": "A0",
"timestamp": "2025-04-14T11:29:21",
"full_text": " \nThe rapid advancement of diffusion models has significantly improved content generation across various modalities, yet video generation still lags in visual quality due to noisy pre-trainin... |
p-10 | 1 | "The generation and manipulation of 3D meshes are pivotal for applications across entertainment, vir(...TRUNCATED) | AI-noInfo | "The generation and manipulation of 3D meshes are pivotal for applications across entertainment, vir(...TRUNCATED) | "The generation and manipulation of 3D meshes are pivotal for applications across entertainment, vir(...TRUNCATED) | generate_content | 75-100% | 4 | null | Male | Some college, no degree | College Student | 18-25 | Maybe | Maybe | Sometimes | "AI-generated text tends to overuse complex vocabulary and/or complex sentence structures while main(...TRUNCATED) | 2 | 2 | 2 | 315 | 20 | 40 | Addition,Substitution | 3 | 93 | -7.77728 | -0.6352 | 75 | 75 | 75 | null | null | null | null | null | null | Accept | [{"time":"12:10:57","step":1,"revision_id":"A0","timestamp":"2025-04-14T12:10:57","full_text":" \nTh(...TRUNCATED) |
p-10 | 2 | "Tool learning has emerged as a promising approach to enhance the applicability of large language mo(...TRUNCATED) | AI-noInfo | "Tool learning has emerged as a promising approach to enhance the applicability of large language mo(...TRUNCATED) | "Tool learning has emerged as a promising approach to enhance the applicability of large language mo(...TRUNCATED) | generate_content | 75-100% | 4 | null | Male | Some college, no degree | College Student | 18-25 | Maybe | Maybe | Sometimes | "AI-generated text tends to overuse complex vocabulary and/or complex sentence structures while main(...TRUNCATED) | 2 | 2 | 2 | 427 | 20 | 40 | Addition,Substitution | 3 | 93 | 21.90132 | 19.37944 | 25 | 75 | 75 | null | null | null | null | null | null | Accept | [{"time":"12:18:40","step":1,"revision_id":"A0","timestamp":"2025-04-14T12:18:40","full_text":" \nTo(...TRUNCATED) |
p-10 | 3 | "Bridging the gap between continuous neural representations and the inherently discrete nature of na(...TRUNCATED) | AI-noInfo | "Bridging the gap between continuous neural representations and the inherently discrete nature of na(...TRUNCATED) | "Bridging the gap between continuous neural representations and the inherently discrete nature of na(...TRUNCATED) | generate_content | 75-100% | 4 | null | Male | Some college, no degree | College Student | 18-25 | Maybe | Maybe | Sometimes | "AI-generated text tends to overuse complex vocabulary and/or complex sentence structures while main(...TRUNCATED) | 2 | 2 | 2 | 444 | 20 | 40 | Addition,Substitution | 3 | 93 | 2.01753 | 6.996558 | 75 | 25 | 75 | null | null | null | null | null | null | Accept | [{"time":"12:23:35","step":1,"revision_id":"A0","timestamp":"2025-04-14T12:23:35","full_text":" \nBr(...TRUNCATED) |
p-100 | 1 | "GUI test case migration is to migrate GUI test cases from a source app to a target app. The key of (...TRUNCATED) | Human-noInfo | "GUI test case migration is to migrate GUI test cases from a source app to a target app. The key of (...TRUNCATED) | "The purpose of GUI test case migration is to migrate GUI test cases from a source app to a target a(...TRUNCATED) | generate_and_edit | 50-75% | null | null | Female | Some college, no degree | Tutor | 18-25 | Maybe | Maybe | Quite often | "AI-generated text uses the (--) quite frequently. It also tends to be verbose without adding much w(...TRUNCATED) | 3 | 2 | 3 | 331 | 20 | 40 | Addition,Deletion,Correction,Substitution,Reordering | 3 | 70 | 34.54576 | 34.60538 | 45 | 70 | null | 80 | 70 | null | "In the method description, the edited abstract uses future tense that is \"we will \", which create(...TRUNCATED) | "This version keeps the extra clarity from the edited abstract (explicit definition of widget matchi(...TRUNCATED) | null | Accept | [{"time":"15:52:55","step":1,"revision_id":"A0","timestamp":"2025-09-02T15:52:55","full_text":" \nGU(...TRUNCATED) |
p-100 | 2 | "Read mapping is a crucial task in computational genomics. Recently, there has been a significant pa(...TRUNCATED) | Human-noInfo | "Read mapping is a crucial task in computational genomics. Recently, there has been a significant pa(...TRUNCATED) | "Read mapping is a crucial task in computational genomics, a growing field dealing with gene therapy(...TRUNCATED) | generate_and_edit | 50-75% | null | null | Female | Some college, no degree | Tutor | 18-25 | Maybe | Maybe | Quite often | "AI-generated text uses the (--) quite frequently. It also tends to be verbose without adding much w(...TRUNCATED) | 3 | 2 | 3 | 416 | 20 | 40 | Addition,Deletion,Correction,Substitution,Reordering | 3 | 75 | 18.80111 | 11.5995 | 71 | 65 | null | 72 | 80 | null | No necessary ideas added to the edited abstract | "I think the edited abstract had more background and detail, making the abstract easier to understan(...TRUNCATED) | null | Accept | [{"time":"16:04:29","step":1,"revision_id":"A0","timestamp":"2025-09-02T16:04:29","full_text":" \nRe(...TRUNCATED) |
p-100 | 3 | "Instance-level explanation of graph neural networks (GNNs) is a well-studied area. These explainers(...TRUNCATED) | Human-noInfo | "Instance-level explanation of graph neural networks (GNNs) is a well-studied area. These explainers(...TRUNCATED) | "Instance-level explanation of graph neural networks (GNNs) is a well-studied area. These explainers(...TRUNCATED) | generate_and_edit | 50-75% | null | null | Female | Some college, no degree | Tutor | 18-25 | Maybe | Maybe | Quite often | "AI-generated text uses the (--) quite frequently. It also tends to be verbose without adding much w(...TRUNCATED) | 3 | 2 | 3 | 1 | 20 | 40 | Addition | 4 | 90 | 18.60375 | 18.60375 | 50 | 50 | 51 | 100 | 100 | 92 | They are the same. | The edited abstract is virtually identical to the original. | I don't see any noticeable difference. | Accept | [{"time":"16:10:57","step":1,"revision_id":"A0","timestamp":"2025-09-02T16:10:57","full_text":" \nIn(...TRUNCATED) |
p-101 | 1 | "Serverless computing has grown rapidly as a new cloud computing paradigm that promises ease-of-mana(...TRUNCATED) | Human-noInfo | "Serverless computing has grown rapidly as a new cloud computing paradigm that promises ease-of-mana(...TRUNCATED) | "Serverless computing has grown rapidly as a new cloud computing paradigm that promises ease-of-mana(...TRUNCATED) | edit_content | 75-100% | null | 5 | Male | A bachelor's degree | Software developer | 26-35 | Maybe | Yes | Rarely | It is not easy to differentiate without using an algorithm to detect | 3 | 6 | 3 | 93 | 30 | 30 | Addition,Deletion,Substitution | 4 | 64 | 9.982094 | 12.66366 | 46 | 50 | 50 | 77 | 91 | 100 | Not much difference between the two abstracts. | Either abstract can be used. Not much difference. | The content of the two abstracts are mostly the same. | Accept | [{"time":"12:10:30","step":1,"revision_id":"A0","timestamp":"2025-09-02T12:10:30","full_text":" \nSe(...TRUNCATED) |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 23