Why are total mentions not equal to recommendation counts?

A mention may express hatred, balanced criticism, or merely an intention to read, so only deduplicated mentions meeting the praise confidence threshold count as recommendations.

What is the difference between Rules-only and Hybrid mode?

Rules-only mode evaluates patterns locally without external network requests, whereas Hybrid mode submits ambiguous or unresolved contexts to a single fixed model call.

Why must the context string contain the entity literally?

Literal matching ensures the sentiment classifier anchors its evaluation to the exact named subject rather than surrounding unrelated clauses.

How does the tool prevent duplicate votes?

Accepted praise is deduplicated using matching commentId and entity pairs, or identical entity and context pairings when commentId is omitted.

How do I apply manual corrections to my results?

Download the unconfirmed correction template, update the classifications, keep the generated id and fingerprint intact, set confirmed to true, and paste the JSON into the corrections input.

Elysia Tools

Navigation

AI Tools

HN mention sentiment: praise, critique and reading intent

Audit sentiment toward each named book or entity in pasted JSON or CSV. Separate praise, critique, reading intent and neutral mentions before counting recommendations.

Details

What this tool helps you do

Paste up to 20 rows with entity and context; optional id and commentId preserve traceability. Context must contain the entity literally. Hybrid mode sends only unresolved contexts to OpenRouter using deepseek/deepseek-v4.1-flash-1017 in one request (45 seconds, no retries or model substitution). Rules-only mode makes no model calls. Scores are heuristic or model-reported, not calibrated accuracy; models can err. Mixed, low-confidence and unresolved mentions never vote automatically. Distinct mentions are not recommendations: accepted praise is deduplicated by commentId/entity or exact entity/context. Aliases are not merged. Download JSON, formula-safe CSV and an unconfirmed correction template. Apply corrections only to the same input by preserving id and fingerprint and setting confirmed: true. This tool does not build a persistent annotation database or train models. Platform and model-provider retention policies still apply.

Execution

Run this tool

Fill in the form, run the tool, and review the result in one place.

Samples

Examples that match this tool

Continue with connected tools and hubs

Mentions: [{"id":"1","entity":"Dune","context":"I loved Dune.","commentId":"101"},{"id":"2","entity":"Dune","context":"I hated Dune.","commentId":"102"},{"id":"3","entity":"Dune","context":"I wanted to read Dune.","commentId":"103"}], Mode: rules, Threshold: 0.85

HN mention sentiment: praise, critique and reading intent

What this tool helps you do

Run this tool

Examples that match this tool

Continue with connected tools and hubs

Learn when to use this tool, what it supports, and how real users apply it.

Key facts

Overview

When to use

How it works

Use cases

Examples

1. Filtering Book Endorsements from Mixed Feedback

2. Auditing Technical Book Mentions from CSV Export

FAQ

CSV Samples

Python Samples

Distributed Tracing Samples

AWS EventBridge Samples

Structured Log Analyzer

PDF Table Extractor to CSV/JSON

Dataset Imbalance Detector & Resampler

Time Series Forecast & Seasonality Analyzer

HN Entity Recommendation Review and Ranking Handoff