recallbench 0.1.0

A universal benchmark harness for AI memory systems
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140

use anyhow::{Context, Result};
use serde::Deserialize;

use crate::traits::BenchmarkDataset;
use crate::types::{BenchmarkQuestion, ConversationSession, Turn};

/// LoCoMo dataset (Snap Research) — long-context conversation memory.
pub struct LoCoMoDataset {
    questions: Vec<BenchmarkQuestion>,
}

#[derive(Debug, Deserialize)]
struct RawLoCoMo {
    #[serde(default)]
    conversation_id: String,
    #[serde(default)]
    conversations: Vec<RawConversation>,
    #[serde(default)]
    qa_pairs: Vec<RawQA>,
}

#[derive(Debug, Deserialize)]
struct RawConversation {
    #[serde(default)]
    session_id: String,
    #[serde(default)]
    date: Option<String>,
    #[serde(default)]
    dialogue: Vec<RawUtterance>,
}

#[derive(Debug, Deserialize)]
struct RawUtterance {
    #[serde(default)]
    speaker: String,
    #[serde(default)]
    text: String,
}

#[derive(Debug, Deserialize)]
struct RawQA {
    #[serde(default)]
    question_id: String,
    #[serde(default)]
    question: String,
    #[serde(default)]
    answer: String,
    #[serde(default, alias = "type")]
    question_type: String,
    #[serde(default)]
    category: Option<String>,
}

impl LoCoMoDataset {
    pub fn from_json(json: &str) -> Result<Self> {
        let raw_data: Vec<RawLoCoMo> = serde_json::from_str(json)
            .context("Failed to parse LoCoMo JSON")?;

        let mut questions = Vec::new();

        for entry in &raw_data {
            let sessions: Vec<ConversationSession> = entry.conversations.iter().map(|conv| {
                ConversationSession {
                    id: conv.session_id.clone(),
                    date: conv.date.clone(),
                    turns: conv.dialogue.iter().map(|u| Turn {
                        role: if u.speaker.to_lowercase().contains("user") { "user".to_string() } else { "assistant".to_string() },
                        content: u.text.clone(),
                    }).collect(),
                }
            }).collect();

            for qa in &entry.qa_pairs {
                let qtype = if !qa.question_type.is_empty() {
                    qa.question_type.clone()
                } else {
                    qa.category.clone().unwrap_or_else(|| "general".to_string())
                };

                questions.push(BenchmarkQuestion {
                    id: qa.question_id.clone(),
                    question_type: qtype,
                    question: qa.question.clone(),
                    ground_truth: vec![qa.answer.clone()],
                    question_date: None,
                    sessions: sessions.clone(),
                    is_abstention: false,
                    metadata: std::collections::HashMap::new(),
                });
            }
        }

        Ok(Self { questions })
    }
}

impl BenchmarkDataset for LoCoMoDataset {
    fn name(&self) -> &str { "locomo" }
    fn variant(&self) -> &str { "default" }
    fn description(&self) -> &str { "LoCoMo (Snap Research) — long-context conversation memory evaluation" }
    fn questions(&self) -> &[BenchmarkQuestion] { &self.questions }
    fn question_types(&self) -> Vec<String> {
        let mut types: Vec<String> = self.questions.iter()
            .map(|q| q.question_type.clone())
            .collect::<std::collections::HashSet<_>>()
            .into_iter().collect();
        types.sort();
        types
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn parse_locomo() {
        let json = r#"[{
            "conversation_id": "conv1",
            "conversations": [{
                "session_id": "s1",
                "date": "2024-01-15",
                "dialogue": [
                    {"speaker": "User", "text": "Hello"},
                    {"speaker": "Assistant", "text": "Hi there"}
                ]
            }],
            "qa_pairs": [{
                "question_id": "q1",
                "question": "What did the user say?",
                "answer": "Hello",
                "type": "factual"
            }]
        }]"#;
        let ds = LoCoMoDataset::from_json(json).unwrap();
        assert_eq!(ds.questions().len(), 1);
        assert_eq!(ds.questions()[0].sessions.len(), 1);
        assert_eq!(ds.questions()[0].sessions[0].turns.len(), 2);
    }
}