1 Introduction
2 Graph Database Consideration
3 Methodology
4 Experimentation Conditions and Results
Cycle 1 | Cycle 2 | |
---|---|---|
Response time (ms) | 5790 | 16978 |
Precision | 0.50 | 0.44 |
Recall | 0.01 | 0.31 |
Root causes | Anomalies distribution |
---|---|
(1) No highlighting performed on the most relevant result | 26% |
(2) The information carried by the table format is not used | 24% |
(3) Information is carried by a term close to the keyword | 19% |
(4) The property carrying the information is close to the keyword | 13% |
(5) Searching through the implicit relationships is impossible | 9% |
(6) The OCR algorithm didn't extract the correct characters | 7% |
(7) Extracting bulleted lists extraction is not correct | 2% |