OMNICHANNEL CONVERSATIONAL SEARCH: MAINTAINING CONTEXT AND CONSISTENCY ACROSS VOICE AND WEB INTERFACES
Bibliographic record
Abstract
This research work presents an omnichannel conversational search platform that provides consistent, context-rich responses via voice and web channels. Current conversational agents address voice and web channels separately, leading to fragmented user experiences and context fragmentation between cross-channel interactions. Our platform, which is built on Dialogflow CX and Google Cloud, enables a single conversational agent to support Telephony Gateway voice calls and Web/Messaging requests. For context and to deliver earthed answers, we employ a Cloud Run tool-router for conversation-aware query rewriting and retrieval with Vertex AI Search/Embeddings with Matching Engine, including span-level citations. Scoped session memory is stored in Firestore with DLP redaction to preserve privacy while permitting shared context between channels. Apigee X enforces operation limits and circuit breakers, Cloud Tasks manages delayed enrichment, and BigQuery with Cloud Trace/Logging supports online and offline evaluation. A channel-consistency controller for synchronized answers across modalities, a handoff linker for state preservation during voice-to-web handoffs, and a latency/cost-aware cache enhancing sub-second FAQ answers are at the core of these. System performance is quantified using metrics like consistency of responses, groundedness, deflection ratio, session continuity after channel switch, customer satisfaction (CSAT/NPS), 95th percentile latency per channel, and cost per solved query. Outcomes demonstrate improved user experience, enhanced reliability of response, and seamless cross-channel interaction, establishing benchmarks for practical use in industrial-scale, multi-channel rollouts. This research has problem is maintaining consistent context and responses across voice and web interfaces, causing fragmented interactions and incoherent omnichannel search experiences. This work highlights the potential of merging AI-driven retrieval and robust context management to enhance omnichannel conversational search systems.
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 | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.005 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.001 | 0.001 |
| 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".