Reconcile

from uiform import UiForm

client = UiForm()

# Multiple extraction results to reconcile
extractions = [
    {
        "title": "Quantum Algorithms in Interstellar Navigation",
        "authors": ["Dr. Stella Voyager", "Dr. Nova Star", "Dr. Lyra Hunter"],
        "year": 2025,
        "keywords": ["quantum computing", "space navigation", "algorithms"]
    },
    {
        "title": "Quantum Algorithms for Interstellar Navigation",
        "authors": ["Dr. S. Voyager", "Dr. N. Star", "Dr. L. Hunter"],
        "year": 2025,
        "keywords": ["quantum algorithms", "interstellar navigation", "space travel"]
    },
    {
        "title": "Application of Quantum Algorithms in Space Navigation",
        "authors": ["Stella Voyager", "Nova Star", "Lyra Hunter"],
        "year": 2025,
        "keywords": ["quantum computing", "navigation", "space exploration"]
    }
]

# Reconcile the different extraction results into a consensus
response = client.consensus.reconcile(
    list_dicts=extractions,
    mode="aligned"
)

consensus_result = response.consensus_dict
consensus_confidence = response.consensus_likelihoods

print(f"Consensus: {consensus_result}")
print(f"Confidence scores: {consensus_confidence}")

{
  "consensus_dict": {
    "authors": [
      "Dr. Stella Voyager",
      "Dr. Nova Star",
      "Dr. Lyra Hunter"
    ],
    "keywords": [
      "quantum computing",
      "space navigation",
      "algorithms"
    ],
    "title": "Quantum Algorithms in Interstellar Navigation",
    "year": 2025
  },
  "likelihoods": {
    "authors": [
      0.76667,
      0.75,
      0.79167
    ],
    "keywords": [
      0.66667,
      0.33333,
      0.33333
    ],
    "title": 0.70897,
    "year": 1.0
  }
}