Direct Answer Threshold Optimization in Dialogue Systems
Bibliographic record
Abstract
The presence of dialogue systems is rising in a wide array of industries. While complex, human-like conversational flow and turn-taking have been the focus of recent research advances, we make the point that direct answers are equally important in human-bot interactions. This is especially true in an information seeking task where prompt, correct answers with minimal back-and-forth are desirable. We define a direct answer as a response given to a user query without requiring further clarifications from the user. To determine whether a direct answer is to be given or not, a threshold is applied to the to the confidence level of the predicted intent; in the case where the confidence is higher than the threshold, the user receives a direct answer. This threshold is often set intuitively or on the basis of a few observations, usually between 50% - 75%. In this paper, we propose a method to estimate this threshold based on the intent classification confidence level combined with several intent volumetrics. The goal of our method is to maximize the number of correct direct responses for as many intents as possible in order to minimize user frustration from unnecessary requests for clarifications. Moreover, our method is applicable in the earlier stages of a dialogue system when real interaction logs are scarce. We show that our method improves the accuracy of directly answered queries by 3 to 14% while maximizing the number of accurately answered intents on two dialogue system datasets of 32 and 152 intents.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.020 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".