How can i measure precision with this example?
How can I calculated precision . I have the total corpus contains 4000 different sentences. If I did semantic search and got 4 sentences are relevant and one is not relevant which k=5. How can I measure it precision here ? i'm so confused and hope get help
the corpus contains 4000 sentences and there 20 sentences talked about the food.
My query is searching for "Sam eat the food " and k=5 gave me the top score using cosine similarity
4 sentences semantically right and there is one is false
here does p1 means k =1 , p2 means k =2 ? so if this is right . will k=1 means that the search for the first sentence has zero result? and for k = 2 , the percentage is 31 ? and so on ?
Precision comes in two ways. Overall precision is the ratio of relevant results among all the search results returned. Precision@k is the ratio of relevant results among the top-k search results returned.
Therefore your precision@5 will be:
precision@5 = relevant_results_at_5 / total_search_results_at_5 = 4/5
- Proximity Search in Search Engines
- How does Elasticsearch do attribute filtering during knn (vector-based) retrieval?
- How can use Haystack to identify the top k sentences that are the closest match to a user query, and then returns the docs containing these sentences?
- Getting the source of information with Langchain
- Wikipedia text download
- Chromadb + Langchain with SentenceTransformerEmbeddingFunction throwing sqlite3 >= 3.35.0 error, despite sqlite3 3.43.0 being available
- Is there a search engine that will give a direct answer?
- Is it possible to use evaluation metrics (like NDCG) as a loss function?
- Is it possible to filter on a non-indexed field with doc_values=true in Elasticsearch
- Retrieving row and column names in R
- Problems understanding nDCG format in pytrec_eval?
- Bitwise operation on a dynamic data structure
- Download small sample of AWS Common Crawl to local machine via http
- Apache Lucene doesn't filter stop words despite the usage of StopAnalyzer and StopFilter
- How can I save file name in the tuples in scala
- Group, map & reduce with two different reducer-operators
- How can i measure precision with this example?
- Keep non-stemmed tokens on Elasticsearch
- How to compute MRR @n?
- Why not use indexwriter to build index in Lucene?
- Fuzzy String Searching with Whoosh in Python
- How to specify two Fields in Lucene QueryParser?
- Can inverted index have multiple words in one entry?
- how to retrieve values from plotly sunburst chart?
- How to build a normalized tf dataframe?
- IR: How do you match documents based on index values and return the document?
- Extract the values of a dataframe that correspond to a single element of another df
- Why restricts Lucene's MoreLikeThis its TermQueries to the field with the highest docFreq?
- How to use metadata for document retrieval using Sentence Transformers?
- Understanding Recall and Precision