Long-term personalized agents are increasingly expected to answer queries over user histories in which preferences, goals, and constraints evolve across interleaved conversations. Existing retrieval-based and structured-memory systems can surface relevant facts, but often return isolated memory units that do not explain which earlier preference was refined, contradicted, or superseded by later evidence. We introduce TransMem, a transition-aware memory retrieval framework that stores preference updates as explicit retrieval units alongside ordinary factual memories. During memory construction, TransMem routes user interaction history into topic-structured memory graphs, detects shift events, and grounds them as transition records. At inference time, TransMem first retrieves factual anchors and then adaptively expands over transition records, producing a compact context of factual anchors and accepted transition records. Experiments on PersonaMem and PersonaBench across two Qwen backbones show that TransMem outperforms representative retrieval-based and structured-memory baselines. Controlled analyses further show that transition records and adaptive expansion contribute beyond factual anchoring, chronological expansion, flat transition retrieval, and fixed-hop traversal. These results suggest that preference-update records are useful retrieval units for long-term personalization.
Figures & tables
Figure 1: Evolving personal preferences: snippets can be stale and temporal facts leave updates implicit, while TransMem retrieves factual anchors with explicit transitions.
Figure 2: Overview of TransMem . Stage 1 (memory construction): ❶ organize messages into topic graphs via the Topic Resonance Router; ❷ compare candidate messages with historical facts to extract transition records; ❸ ground transition records into structured personal memory shown in the center. Stage 2 (inference): retrieve factual anchors and adaptively expand over transition records to assemble the final context.
Method
Revisit Reasons
Track Evolution
Latest Prefs
Aligned Recs
New Scenario
Shared Facts
New Ideas
Overall
Qwen3-8B
ZeroMem
57.41
55.00
29.41
29.63
18.37
35.40
0 9.09
34.53
NaiveRAG
68.52
54.00
58.82
38.89
51.02
54.87
18.18
47.37
RF-Mem
53.70
56.00
35.29
61.11
42.86
57.52
19.32
47.79
Zep
68.52
55.00
52.94
66.67
61.22
54.87
22.73
52.42
A-Mem
75.93
61.00
70.59
62.96
44.90
53.98
23.86
53.05
Mem0
66.67
51.00
64.71
61.11
48.98
51.33
10.23
46.74
Table 1: PersonaMem results in the 32k context regime. Best results in bold; second-best underlined.
Method
Basic Info
Pref Easy
Pref Hard
Social
Overall
Qwen3-8B
NaiveRAG
13.96
21.29
24.48
15.65
17.88
RF-Mem
18.63
23.80
18.60
15.88
18.58
A-Mem
16.71
21.54
26.71
21.82
20.21
LightMem
19.37
23.38
29.05
23.21
22.43
TransMem (ours)
23.80
25.14
30.16
22.00
24.67
Qwen3.5-27B
NaiveRAG
14.31
23.11
22.93
16.08
18.07
Table 3: PersonaBench results (lexical F1, %). Best results in bold; second-best underlined.
Figure 3: Ablation results comparing four retrieval variants across backbone and context settings. Facts Only : factual retrieval without transition edges; Intra Transitions : intra-topic transition edges only; Inter Transitions : inter-topic transition edges only; Full Transitions : TransMem with both edge types.
Figure 4: Left : Inference latency and token usage versus overall accuracy in the 32k-8B setting; bubble area is proportional to inference token count. Right : Comparison of TransMem with chronological expansion ( ChronoExpand ), flat transition retrieval ( FlatTrans ), and fixed-hop transition expansion.
Figure 5: Per-category accuracy and adaptive retrieval depth on PersonaMem-128k (Qwen3-8B). Bars compare Facts Only with Full Transitions ; the right axis shows average expansion depth.
Appendix figures & tables18 assets
Supplementary material from the paper’s appendix.
Appendix
Statistic
32k
128k
User cases
37
55
Questions
475
2,161
Avg. history tokens / user
26,592.51
121,346.55
Avg. messages / user
174.65
761.91
Appendix
Table 4: PersonaMem evaluation-set statistics after filtering. Token counts use tiktoken with cl100k_base .
Aspect
Evaluates
Revisit Reasons
Recall the reason or event that caused a user preference to change.
Track Evolution
Recover how a user preference changes across time rather than only the latest state.
Latest Prefs
Recognize the most recent preference expressed in the user history.
Aligned Recs
Recommend items or activities aligned with the user’s current preferences.
New Scenario
Apply user preferences to a new situation not directly seen in the history.
Shared Facts
Recall static events, activities, or interests the user previously shared.
Appendix
Table 5: Evaluation aspects in PersonaMem.
32k
128k
Aspect
Before
After
Before
After
Revisit Reasons
99 (16.81%)
54 (11.37%)
237 (9.69%)
158 (7.31%)
Track Evolution
139 (23.60%)
100 (21.05%)
300 (12.26%)
220 (10.18%)
Latest Prefs
17 (2.89%)
17 (3.58%)
772 (31.56%)
689 (31.88%)
Aligned Recs
55 (9.34%)
54 (11.37%)
320 (13.08%)
310 (14.35%)
New Scenario
57 (9.68%)
49 (10.32%)
195 (7.97%)
171 (7.91%)
Appendix
Table 6: PersonaMem query-type distribution before and after filtering. Each cell reports the count and its percentage within the corresponding split.
Method
Input
Output
Total
Mem0
434,977
121,250
556,227
A-Mem
867,453
82,513
949,966
TransMem (ours)
620,976
101,201
722,177
Appendix
Table 7: Average per-user token cost for offline memory construction (PersonaMem 32k, Qwen3.5-27B).
Type
Old state
New state
Update rationale
Valid time
Music preference shift
Engaging with the quirky and humorous side of music
Focusing on emotional and reflective aspects of music
The user moved from lighthearted storytelling to a more introspective music experience.
Jargon-heavy material caused confusion, while informal mentoring worked better.
2018-09-14
Appendix
Table 8: Examples of transition records produced by the shift detector.
Context
Model
Revisit Reasons
Track Evo
Latest Prefs
Aligned Recs
New Scenario
Shared Facts
New Ideas
Overall
32k
Qwen3-8B
62.96
64.00
47.06
42.59
53.06
46.02
10.23
45.47
Qwen3.5-27B
90.74
77.00
58.82
61.11
77.55
69.03
27.27
65.05
128k
Qwen3-8B
— (exceeds backbone context limit)
Qwen3.5-27B
89.24
77.73
60.81
52.58
76.61
71.43
33.33
59.60
Appendix
Table 9: FullText results across all configurations. The 128k + Qwen3-8B entry is unavailable because the full history exceeds the model’s context limit.
Topic
n
Qwen3-8B
Qwen3.5-27B
Food
2
100.00
100.00
Home
2
100.00
100.00
Medical
14
0 92.86
0 92.86
Legal
28
0 64.29
0 75.00
Travel
60
0 60.00
0 70.00
Study
29
0 58.62
0 62.07
Appendix
Table 10: Per-task accuracy of TransMem in the 32k context regime. n = evaluation queries per topic.
Topic
n
Qwen3-8B
Qwen3.5-27B
Home
111
58.56
63.96
Study
131
58.02
74.81
Family
117
55.56
70.94
Travel
126
53.97
65.87
Music
38
52.63
73.68
Medical
160
51.25
73.12
Appendix
Table 11: Per-task accuracy of TransMem in the 128k context regime. n = evaluation queries per topic.
Window h
Overall Acc. (%)
1
55.64
2
56.03
3
53.78
Appendix
Table 12: ChronoExpand overall accuracy on the diagnostic subset (20 users, 257 queries) for different chronological window sizes.
Category
Facts Only
Full
Δ
Avg Hops
Revisit
46.84
51.90
+5.06
2.17
Track Evo
51.82
55.00
+3.18
2.51
Latest Prefs
53.12
56.75
+3.63
1.90
Aligned Recs
51.29
51.29
+ 0.00
1.72
New Scenario
47.37
53.22
+5.85
1.81
Shared Facts
57.79
62.99
+5.20
1.89
Appendix
Table 13: Per-category Facts Only vs. Full TransMem with average adaptive hop depth (128k, Qwen3-8B).
Figure 6: Per-category comparison of Facts Only versus Full TransMem in the 128k setting (Qwen3.5-27B). Bars show accuracy; the right axis shows the average adaptive hop depth of the full model.
Category
Facts Only
Full
Δ
Avg Hops
Revisit
70.25
70.89
+0.64
1.92
Track Evo
60.00
62.27
+2.27
2.24
Latest Prefs
67.63
69.23
+1.60
1.67
Aligned Recs
66.13
67.74
+1.61
1.52
New Scenario
71.93
73.68
+1.75
1.63
Shared Facts
81.82
85.06
+3.24
1.63
Appendix
Table 14: Per-category Facts Only vs. Full TransMem with average adaptive hop depth (128k, Qwen3.5-27B).
Configuration
Topic Router
Trans. Edges
Acc (%)
Single topic, no transitions, no router
✗
✗
51.75
Single topic + transition edges (intra only)
✗
✓
54.86
Multi-topic router, no transitions
✓
✗
55.64
Full TransMem (multi-topic + transitions)
✓
✓
57.59
Appendix
Table 15: Validation of topic routing and transition edges (20-user diagnostic subset, 257 queries). Overall accuracy (%).
τmacro
τmicro
κ
Acc (%)
0.65
0.25
5
56.81
0.70
0.25
5
54.47
0.65
0.20
10
56.42
0.65
0.25
20
56.81
Appendix
Table 16: Construction hyperparameter sensitivity (20-user diagnostic subset).
Record type
Sampled
Valid rate
Intra-topic
100
98.0%
Inter-topic
100
99.0%
Total
200
98.5%
Appendix
Table 17: LLM-judged quality of stored transition records.
Figure 7: Sensitivity to the factual-anchor budget k . (a) Qwen3-8B; (b) Qwen3.5-27B. The adopted value is marked with a star.
Figure 8: Sensitivity to the trajectory-retention factor γ . Two y -axes are used because the two backbones differ by roughly 17 pp on this subset. The adopted value for each backbone is marked with a star.
Backbone
(τc,δc)
Acc (%)
Qwen3-8B
(0.40,0.10)
57.59
(0.50,0.10)
56.81
(0.50,0.20)
54.86
(0.40,0.20)
52.14
Qwen3.5-27B
(0.40,0.10)
72.37
(0.50,0.20)
71.98
Appendix
Table 18: Continuation/stopping sensitivity (Equation 12 ). Values are overall accuracy (%). Best configuration in bold.