Elasticsearch:Use Elasticsearch Machine Learning to implement intelligent question answering

Preparations

Step 1: Index data

1. Use an ingestion pipeline to configure the models in an index to complete text conversion during data writing.

   Note

   You must use models to convert existing data in the index into vectors. For more information, see Ingest documents.

   PUT _ingest/pipeline/text-embedding-pipeline
   {
     "processors": [
       {
         "inference": {
           "model_id": "models--thenlper--gte-large-zh",
           "target_field": "text_embedding",
           "field_map": {
             "context": "text_field"
           }
         }
       }
     ]
   }
   
   #####Configure mappings for the index and specify an ingestion pipeline.
   PUT question_answering
   {
     "settings": {
       "index": {
             "number_of_shards": 2,
             "number_of_replicas": 1,
             "default_pipeline": "text-embedding-pipeline"
         }
     }, 
     "mappings": {
       "properties": {
         "text_embedding.predicted_value": {
           "type": "dense_vector",
           "dims": 1024,
           "index": true,
           "similarity": "l2_norm"
         },
         "context": {
           "type": "text",
           "fields": {
             "keyword": {
               "type": "keyword",
               "ignore_above": 256
             }
           }
         },
         "title": {
           "type": "text",
           "fields": {
             "keyword": {
               "type": "keyword",
               "ignore_above": 256
             }
           }
         }
       }
     }
   }

2. Use a Python environment to import the prepared dataset. In this example, Python 3.10 is used.

   # -*- coding:utf-8 -*-
   import json
   from pprint import pprint
   import os
   import time
   from elasticsearch import helpers
   import time
   from elasticsearch import Elasticsearch
   
   elastic_user="elastic"
   elastic_password="****"
   elastic_endpoint="es-cn-lbj3l7erv0009****.elasticsearch.aliyuncs.com"
   url = f"http://{elastic_user}:{elastic_password}@{elastic_endpoint}:9200"
   es = Elasticsearch(url)
   print(es.info())
   with open('/model/extracted_data.json') as f:
    data_json = json.load(f)
   
   # Prepare the documents to be indexed
   documents = []
   for doc in data_json:
    documents.append({
    "_index": "question_answering",
    "_source": doc,
    })
   
   # Use helpers.bulk to index
   helpers.bulk(es, documents)
   
   print("Done indexing documents into `question_answering` index!")
   time.sleep(5)

Step 2: Retrieve data in the Kibana console

Use the question "What are the use scenarios of Alibaba Cloud Elasticsearch?" as an example to perform a retrieval in the Kibana console of the Elasticsearch cluster.

1. Click the icon in the upper-left corner of the Kibana console. In the left-side navigation pane, choose Management > Dev Tools.

2. Call the inference API operation for Alibaba Cloud Machine Learning to convert documents into vector data.

   Elasticsearch clusters of V8.7 or later

   If you use an Elasticsearch cluster of V8.7 or later, you can use the query_vector_builder configuration object to build a query vector.

   GET question_answering/_search
   {
     "_source": ["context","title"],
     "knn": {
       "field": "text_embedding.predicted_value",
       "k": 5,
       "num_candidates": 10,
       "query_vector_builder": {
         "text_embedding": { 
           "model_id": "models--thenlper--gte-large-zh", 
           "model_text": "What are the use scenarios of Alibaba Cloud Elasticsearch?" 
         }
       }
     }
   }

   The following result is returned. The result shows that the first five documents are highly correlative to Elasticsearch and the first document mainly introduces Alibaba Cloud Elasticsearch.

   {
     "took": 30,
     "timed_out": false,
     "_shards": {
       "total": 2,
       "successful": 2,
       "skipped": 0,
       "failed": 0
     },
     "hits": {
       "total": {
         "value": 5,
         "relation": "eq"
       },
       "max_score": 0.003581697,
       "hits": [
         {
           "_index": "question_answering",
           "_id": "evdbaI0BU_1of7kX5_Hn",
           "_score": 0.003581697,
           "_source": {
             "context": "Alibaba Cloud Elasticsearch is a fully managed cloud service that is developed based on open source Elasticsearch. This service is fully compatible with the features provided by open source Elasticsearch. It is out-of-the-box and supports the pay-as-you-go billing method. In addition to Elastic Stack components such as Elasticsearch, Logstash, Kibana, and Beats, Alibaba Cloud provides the X-Pack plug-in (advanced features of the open source Elasticsearch Platinum edition) free of charge together with Elastic. X-Pack is integrated into Kibana to provide features, such as security, alerting, monitoring, and machine learning. It also provides SQL capabilities. Alibaba Cloud Elasticsearch is widely used in scenarios such as real-time log analysis and processing, information retrieval, multidimensional data queries, and statistical data analytics.",
             "title": "What is Alibaba Cloud Elasticsearch?"
           }
         },
         {
           "_index": "question_answering",
           "_id": "2VBdaI0Bg5EzGjIAN-Tx",
           "_score": 0.003581697,
           "_source": {
             "context": "Alibaba Cloud Elasticsearch is a fully managed cloud service that is developed based on open source Elasticsearch. This service is fully compatible with the features provided by open source Elasticsearch. It is out-of-the-box and supports the pay-as-you-go billing method. In addition to Elastic Stack components such as Elasticsearch, Logstash, Kibana, and Beats, Alibaba Cloud provides the X-Pack plug-in (advanced features of the open source Elasticsearch Platinum edition) free of charge together with Elastic. X-Pack is integrated into Kibana to provide features, such as security, alerting, monitoring, and machine learning. It also provides SQL capabilities. Alibaba Cloud Elasticsearch is widely used in scenarios such as real-time log analysis and processing, information retrieval, multidimensional data queries, and statistical data analytics.",
             "title": "What is Alibaba Cloud Elasticsearch?"
           }
         },
         {
           "_index": "question_answering",
           "_id": "t49gaI0BNqB2ciGcDbmF",
           "_score": 0.003581697,
           "_source": {
             "context": "Alibaba Cloud Elasticsearch is a fully managed cloud service that is developed based on open source Elasticsearch. This service is fully compatible with the features provided by open source Elasticsearch. It is out-of-the-box and supports the pay-as-you-go billing method. In addition to Elastic Stack components such as Elasticsearch, Logstash, Kibana, and Beats, Alibaba Cloud provides the X-Pack plug-in (advanced features of the open source Elasticsearch Platinum edition) free of charge together with Elastic. X-Pack is integrated into Kibana to provide features, such as security, alerting, monitoring, and machine learning. It also provides SQL capabilities. Alibaba Cloud Elasticsearch is widely used in scenarios such as real-time log analysis and processing, information retrieval, multidimensional data queries, and statistical data analytics.",
             "title": "What is Alibaba Cloud Elasticsearch?"
           }
         },
         {
           "_index": "question_answering",
           "_id": "efdbaI0BU_1of7kX5_Hn",
           "_score": 0.0027631863,
           "_source": {
             "context": "For information about how to log on to the Kibana console, see the "Log on to the Kibana console" topic. The username that you can use to log on to the Kibana console is elastic. The password is the one that you specified when you create your Elasticsearch cluster. If you forget your password, you can reset it. For information about the procedure and precautions for resetting the password, see the "Reset the access password for an Elasticsearch cluster" topic.",
             "title": "How do I log on to the Kibana console? What are the username and password that I can use to log on to the Kibana console?"
           }
         },
         {
           "_index": "question_answering",
           "_id": "2FBdaI0Bg5EzGjIAN-Tx",
           "_score": 0.0027631863,
           "_source": {
             "context": "For information about how to log on to the Kibana console, see the "Log on to the Kibana console" topic. The username that you can use to log on to the Kibana console is elastic. The password is the one that you specified when you create your Elasticsearch cluster. If you forget your password, you can reset it. For information about the procedure and precautions for resetting the password, see the "Reset the access password for an Elasticsearch cluster" topic.",
             "title": "How do I log on to the Kibana console? What are the username and password that I can use to log on to the Kibana console?"
           }
         }
       ]
     }
   }

   Elasticsearch clusters earlier than V8.7

   1. Call the inference API operation for Elasticsearch Machine Learning.

      POST /_ml/trained_models/thenlper__gte-large-zh/_infer
      {
        "docs":[{"text_field": "What are the use scenarios of Alibaba Cloud Elasticsearch?"}]
      }

      The following result is returned.

      

   2. Copy the vector data and call the kNN search API to search for similar documents in the index.

      GET question_answering/_search
      { 
        "_source": ["context","title"],
        "knn": {
          "field": "text_embedding.predicted_value",
          "k": 5,
          "num_candidates": 10,
          "query_vector": [
            0.2767493426799774,
            0.05577810853719711,
            0.2760164141654968,
            -0.9484721422195435,
            ..............
            0.8358230590820312,
            0.6053569316864014,
            -0.5380803942680359
          ]
        }
      }

3. Copy the value of the context parameter for the first document and call the question answering model to implement question answering.

   POST /_ml/trained_models/models--luhua--chinese_pretrain_mrc_macbert_large/_infer
   {
     "docs":[{"text_field": "Alibaba Cloud Elasticsearch is a fully managed cloud service that is developed based on open source Elasticsearch. This service is fully compatible with the features provided by open source Elasticsearch. It is out-of-the-box and supports the pay-as-you-go billing method. In addition to Elastic Stack components such as Elasticsearch, Logstash, Kibana, and Beats, Alibaba Cloud provides the X-Pack plug-in (advanced features of the open source Elasticsearch Platinum edition) free of charge together with Elastic. X-Pack is integrated into Kibana to provide features, such as security, alerting, monitoring, and machine learning. It also provides SQL capabilities. Alibaba Cloud Elasticsearch is widely used in scenarios such as real-time log analysis and processing, information retrieval, multidimensional data queries, and statistical data analytics."}],
     "inference_config": {"question_answering": {"question": "What are the use scenarios of Alibaba Cloud Elasticsearch?"}}
   }

   The following result is returned:

   {
     "inference_results": [
       {
         "predicted_value": "Real-time log analysis and processing",
         "start_offset": 220,
         "end_offset": 228,
         "prediction_probability": 0.014678373776954107
       }
     ]
   }

Step 3: Simulate a question answering scenario

Run the Python3 command in the ECS instance to load the Python environment. Then, run the following command to simulate a question answering scenario:

# -*- coding:utf-8  -*-
import json
from pprint import pprint
import os
import time
from elasticsearch import helpers
import time
from elasticsearch import Elasticsearch

elastic_user="elastic"
elastic_password="Es123456"
elastic_endpoint="es-cn-lbj3l7erv0009bz81.public.elasticsearch.aliyuncs.com"
url = f"http://{elastic_user}:{elastic_password}@{elastic_endpoint}:9200"
es = Elasticsearch(url)

question = input("Enter your question: ")

query_args = {
    "field": "text_embedding.predicted_value",
    "k": 5,
    "num_candidates": 10,
    "query_vector_builder": {
        "text_embedding": {
            "model_id": "models--thenlper--gte-large-zh",
            "model_text": question
        }
    }
}

result=es.search(index="question_answering",source=["context","title"],knn=query_args)
# print(result["hits"]["hits"])
concatenated_content = ""
for hit in result["hits"]["hits"]:
    context=hit["_source"]["context"]
    concatenated_content += context+" "

#print(concatenated_content)
docs=[{"text_field": concatenated_content}]
inference_config= {"question_answering": {"question": question}}

response=es.ml.infer_trained_model(model_id="models--luhua--chinese_pretrain_mrc_macbert_large",docs=docs,inference_config=inference_config)
print("Inference result: ")
print(response["inference_results"][0]["predicted_value"])

The following result is returned:

Enter your question:
Where is the location of Zhangjiagang High-Speed Railway Station?
Inference result:
Zhangjiagang High-speed Railway Station is located on the east side of the new National Highway 204, north of Renmin Road in Tangqiao Town.