{"dataset_uri":"https://raw.githubusercontent.com/demoPlz/mini-template/main/studio/dataset.csv","field_refinements":[{"field":"Conference","semantic_type":"Category","separator":null,"date_format":null,"is_cryptic_code":false,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"463650e05f7fcfb7"}},{"field":"Year","semantic_type":"Year","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"abf6e54c35745021"}},{"field":"Title","semantic_type":"Text","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"dd802219497bfe2c"}},{"field":"DOI","semantic_type":"Identifier","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"d313fec31b84f869"}},{"field":"Link","semantic_type":"URL","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"587e21f3c0362e6a"}},{"field":"FirstPage","semantic_type":"Numeric","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"28df9f1ece147d6d"}},{"field":"LastPage","semantic_type":"Numeric","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"9901463e5ef4dcd2"}},{"field":"PaperType","semantic_type":"Category","separator":null,"date_format":null,"is_cryptic_code":true,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"d925258915ad2e9f"}},{"field":"Abstract","semantic_type":"Text","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"af59095bf0729951"}},{"field":"AuthorNames-Deduped","semantic_type":"Name","separator":";","date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"0c21543c7f9c7dca"}},{"field":"AuthorNames","semantic_type":"Name","separator":";","date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"73028ab15cd647a9"}},{"field":"AuthorAffiliation","semantic_type":"Text","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"6fb08ebc30a4f995"}},{"field":"InternalReferences","semantic_type":"Identifier","separator":";","date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"b641a71f1e33da9c"}},{"field":"AuthorKeywords","semantic_type":"Category","separator":",","date_format":null,"is_cryptic_code":false,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"e1b04cce1a4ed30d"}},{"field":"AminerCitationCount","semantic_type":"Numeric","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"bbbe6a9cc1fb84f3"}},{"field":"CitationCount_CrossRef","semantic_type":"Numeric","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"b4735c1721253501"}},{"field":"PubsCited_CrossRef","semantic_type":"Numeric","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"326deb3c4ff02388"}},{"field":"Downloads_Xplore","semantic_type":"Numeric","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"0a6bd1a6beaf8a6a"}},{"field":"Award","semantic_type":"Category","separator":";","date_format":null,"is_cryptic_code":true,"boolean_truthy_value":null,"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"73fcec55fea104b2"}},{"field":"GraphicsReplicabilityStamp","semantic_type":"Boolean","separator":null,"date_format":null,"is_cryptic_code":null,"boolean_truthy_value":"X","__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"7d0b0cf93cd5c28b"}}],"field_expansions":[{"field":"PaperType","mapping":{"C":"Conference Paper","J":"Journal Paper","M":"Miscellaneous"},"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"3fbcd61456b060bf"}},{"field":"Award","mapping":{"ba":"Best Application Award","bcs":"Best Conference Short Paper","bp":"Best Paper Award","hm":"Honorable Mention","tt":"Test of Time Award"},"__metadata__":{"trace_id":"b088b1f30dfbe43a4a3995f28cff946f","observation_id":"4f6ac188265b5e92"}}],"goals":{"Track the total publication count and average citation count (using AminerCitationCount) over Year, segmented by Conference, to show the growth of the field and impact within different venues.":{"goal":"Track the total publication count and average citation count (using AminerCitationCount) over Year, segmented by Conference, to show the growth of the field and impact within different venues.","sql":"-- Calculate annual publication counts and average citations by conference\nSELECT\n    MAKE_DATE(\"Year\", 1, 1) AS publication_year,\n    \"Conference\" AS conference,\n    COUNT(\"DOI\") AS publication_count,\n    -- Calculate average citations, excluding NULLs to prevent skewed results\n    AVG(\"AminerCitationCount\") FILTER (WHERE \"AminerCitationCount\" IS NOT NULL) AS average_aminer_citations\nFROM\n    $tablename\nWHERE\n    \"Year\" IS NOT NULL\n    AND \"Conference\" IS NOT NULL\nGROUP BY\n    publication_year,\n    conference\nORDER BY\n    publication_year ASC,\n    conference ASC;","fields":{"summary":"A summary of the total number of publications and their average Aminer citation count, aggregated annually for each conference. This view tracks the growth and impact of research across different publication venues over time.","fields":[{"name":"publication_year","label":"Publication Year","description":"Derived from Year: The publication year, converted to a DATE format representing January 1st of that year using MAKE_DATE(Year, 1, 1).","semantic_type":"Year","role":"primary"},{"name":"conference","label":"Conference","description":"Original field Conference used without modification.","semantic_type":"Category","role":"segment"},{"name":"publication_count","label":"Publication Count","description":"Derived from DOI: The total count of publications, aggregated by publication_year and conference.","semantic_type":"Numeric","role":"primary"},{"name":"average_aminer_citations","label":"Average Aminer Citations","description":"Derived from AminerCitationCount: The average number of citations from the Aminer database, aggregated by publication_year and conference. Publications with no citation data (NULL) are excluded from the average calculation.","semantic_type":"Numeric","role":"detail"}]}},"Show the distribution of PaperType over Year, to illustrate how publication types have evolved or remained consistent over time.":{"goal":"Show the distribution of PaperType over Year, to illustrate how publication types have evolved or remained consistent over time.","sql":"-- Calculate the count of publications for each paper type per year\nSELECT\n    MAKE_DATE(\"Year\", 1, 1) AS publication_year,\n    \"PaperType\" AS paper_type,\n    COUNT(*) AS number_of_papers\nFROM $tablename\nWHERE\n    \"PaperType\" IS NOT NULL\nGROUP BY\n    1, 2\nORDER BY\n    publication_year ASC,\n    number_of_papers DESC;","fields":{"summary":"A summary of the number of publications for each paper type, tracked annually, to show trends in publication formats over time.","fields":[{"name":"publication_year","label":"Publication Year","description":"Derived from Year: The publication year, converted to a DATE format representing January 1st of the given year using MAKE_DATE(\"Year\", 1, 1).","semantic_type":"DateTime","role":"primary"},{"name":"paper_type","label":"Paper Type","description":"Original field PaperType used without modification. Records where the paper type is NULL have been excluded.","semantic_type":"Category","role":"segment"},{"name":"number_of_papers","label":"Number of Papers","description":"Derived from table records: A count of publications, aggregated by publication year and paper type.","semantic_type":"Numeric","role":"primary"}]}},"Analyze the correlation between individual AminerCitationCount and Downloads_Xplore, colored by PaperType and faceted by whether the publication received an Award (derived: HasAward), to understand the relationship between citations and downloads and influencing factors.":{"goal":"Analyze the correlation between individual AminerCitationCount and Downloads_Xplore, colored by PaperType and faceted by whether the publication received an Award (derived: HasAward), to understand the relationship between citations and downloads and influencing factors.","sql":"-- This query selects individual publication records to analyze the correlation\n-- between citation counts and downloads, including contextual attributes for detailed exploration.\nSELECT\n    TRIM(\"Title\") AS title,\n    \"DOI\" AS doi,\n    \"AminerCitationCount\" AS aminer_citation_count,\n    \"Downloads_Xplore\" AS downloads_xplore,\n    COALESCE(NULLIF(TRIM(\"PaperType\"), ''), 'Unknown') AS paper_type,\n    -- Derive a boolean category indicating if a publication has received an award.\n    CASE\n        WHEN len(\"Award\") > 0 THEN 'Has Award'\n        ELSE 'No Award'\n    END AS has_award,\n    -- Detail fields for context and cross-filtering\n    \"Conference\" AS conference,\n    MAKE_DATE(\"Year\", 1, 1) AS publication_year,\n    CASE\n        WHEN \"GraphicsReplicabilityStamp\" THEN 'Replicable'\n        ELSE 'Not Replicable'\n    END AS graphics_replicability,\n    \"Link\" AS url\nFROM\n    $tablename\nWHERE\n    \"AminerCitationCount\" IS NOT NULL\n    AND \"Downloads_Xplore\" IS NOT NULL","fields":{"summary":"A detailed view of individual academic publications, showing the relationship between Aminer citation counts and IEEE Xplore downloads. Data is segmented by paper type and whether the publication received an award, with additional details for interactive exploration.","fields":[{"name":"title","label":"Title","description":"Derived from Title: Direct selection with TRIM() applied.","semantic_type":"Name","role":"detail"},{"name":"doi","label":"DOI","description":"Original field DOI used without modification.","semantic_type":"Identifier","role":"detail"},{"name":"aminer_citation_count","label":"Aminer Citations","description":"Original field AminerCitationCount used without modification.","semantic_type":"Numeric","role":"primary"},{"name":"downloads_xplore","label":"Xplore Downloads","description":"Original field Downloads_Xplore used without modification.","semantic_type":"Numeric","role":"primary"},{"name":"paper_type","label":"Paper Type","description":"Derived from PaperType: Cleaned using TRIM() and COALESCE to handle empty strings and NULLs, defaulting to 'Unknown'.","semantic_type":"Category","role":"segment"},{"name":"has_award","label":"Has Award","description":"Derived from Award: Calculated using a CASE statement to check if the Award list is empty (len(\"Award\") > 0).","semantic_type":"Boolean","role":"facet"},{"name":"conference","label":"Conference","description":"Original field Conference used without modification.","semantic_type":"Category","role":"detail"},{"name":"publication_year","label":"Publication Year","description":"Derived from Year: Converted to a DATE type using MAKE_DATE(\"Year\", 1, 1).","semantic_type":"DateTime","role":"detail"},{"name":"graphics_replicability","label":"Graphics Replicability","description":"Derived from GraphicsReplicabilityStamp: Converted boolean to human-readable labels 'Replicable' or 'Not Replicable' using a CASE statement.","semantic_type":"Boolean","role":"detail"},{"name":"url","label":"URL","description":"Original field Link used without modification.","semantic_type":"URL","role":"detail"}]}},"Examine the relationship between individual publication Year and PubsCited_CrossRef, segmented by Conference, to see if citation practices vary across time or venues.":{"goal":"Examine the relationship between individual publication Year and PubsCited_CrossRef, segmented by Conference, to see if citation practices vary across time or venues.","sql":"-- Select individual publication data for scatter plot analysis\nSELECT\n    \"Title\" AS title,\n    \"DOI\" AS doi,\n    -- Convert Year to a proper DATE format for temporal analysis\n    MAKE_DATE(\"Year\", 1, 1) AS publication_date,\n    \"PubsCited_CrossRef\" AS publications_cited,\n    COALESCE(NULLIF(TRIM(\"Conference\"), ''), 'Unknown') AS conference,\n    \"Link\" AS url,\n    COALESCE(NULLIF(TRIM(\"PaperType\"), ''), 'Unknown') AS paper_type,\n    CASE\n        WHEN \"GraphicsReplicabilityStamp\" THEN 'Replicable'\n        ELSE 'Not Replicable'\n    END AS graphics_replicability,\n    CASE\n        WHEN len(\"Award\") > 0 THEN 'Award Winner'\n        ELSE 'No Award'\n    END AS award_status,\n    \"AminerCitationCount\" AS aminer_citations,\n    \"CitationCount_CrossRef\" AS crossref_citations\nFROM\n    $tablename\nWHERE\n    \"PubsCited_CrossRef\" IS NOT NULL\n    AND \"Year\" IS NOT NULL;","fields":{"summary":"A detailed view of individual academic publications, showing the number of works they cited against their publication year. Each point represents a single paper, segmented by conference, to analyze citation patterns over time and across different venues.","fields":[{"name":"title","label":"Title","description":"Original field Title used without modification.","semantic_type":"Name","role":"detail"},{"name":"doi","label":"DOI","description":"Original field DOI used without modification.","semantic_type":"Identifier","role":"detail"},{"name":"publication_date","label":"Publication Date","description":"Derived from Year: Converted to a DATE format using MAKE_DATE(Year, 1, 1).","semantic_type":"DateTime","role":"primary"},{"name":"publications_cited","label":"Publications Cited","description":"Original field PubsCited_CrossRef used without modification.","semantic_type":"Numeric","role":"primary"},{"name":"conference","label":"Conference","description":"Derived from Conference: Cleaned using TRIM() and replaced NULL or empty values with 'Unknown'.","semantic_type":"Category","role":"segment"},{"name":"url","label":"URL","description":"Original field Link used without modification.","semantic_type":"URL","role":"detail"},{"name":"paper_type","label":"Paper Type","description":"Derived from PaperType: Cleaned using TRIM() and replaced NULL or empty values with 'Unknown'.","semantic_type":"Category","role":"detail"},{"name":"graphics_replicability","label":"Graphics Replicability","description":"Derived from GraphicsReplicabilityStamp: Converted boolean to human-readable labels ('Replicable', 'Not Replicable').","semantic_type":"Category","role":"detail"},{"name":"award_status","label":"Award Status","description":"Derived from Award: Categorized based on whether the Award list is empty ('No Award') or contains entries ('Award Winner').","semantic_type":"Category","role":"detail"},{"name":"aminer_citations","label":"Aminer Citations","description":"Original field AminerCitationCount used without modification.","semantic_type":"Numeric","role":"detail"},{"name":"crossref_citations","label":"CrossRef Citations","description":"Original field CitationCount_CrossRef used without modification.","semantic_type":"Numeric","role":"detail"}]}},"Rank the top 10 Conferences by total publication count and average CitationCount_CrossRef, to identify the most prolific and impactful publication venues.":{"goal":"Rank the top 10 Conferences by total publication count and average CitationCount_CrossRef, to identify the most prolific and impactful publication venues.","sql":"-- Rank the top 10 conferences by publication count and calculate their average citation count\nSELECT\n  \"Conference\" AS conference,\n  COUNT(\"DOI\") AS total_publications,\n  AVG(\"CitationCount_CrossRef\") FILTER (\n    WHERE\n      \"CitationCount_CrossRef\" IS NOT NULL\n  ) AS average_citations\nFROM $tablename\nWHERE\n  NULLIF(TRIM(\"Conference\"), '') IS NOT NULL\nGROUP BY\n  \"Conference\"\nORDER BY\n  total_publications DESC\nLIMIT 10;","fields":{"summary":"A ranked list of the top 10 most prolific and impactful conferences, based on their total number of publications and average CrossRef citation counts.","fields":[{"name":"conference","label":"Conference","description":"Derived from Conference: Direct selection after filtering out null or empty values.","semantic_type":"Name","role":"primary"},{"name":"total_publications","label":"Total Publications","description":"Derived from DOI: Count aggregated by Conference.","semantic_type":"Numeric","role":"primary"},{"name":"average_citations","label":"Average Citations","description":"Derived from CitationCount_CrossRef: Average aggregated by Conference, excluding null values.","semantic_type":"Numeric","role":"segment"}]}},"Identify the top 10 AuthorNames-Deduped by their total publication count and sum of AminerCitationCount, to highlight influential researchers.":{"goal":"Identify the top 10 AuthorNames-Deduped by their total publication count and sum of AminerCitationCount, to highlight influential researchers.","sql":"WITH AuthorPublications AS (\n  -- Unnest the list of authors to create a row for each author per publication\n  SELECT\n    UNNEST(\"AuthorNames-Deduped\") AS author_name,\n    AminerCitationCount\n  FROM $tablename\n)\nSELECT\n  author_name,\n  COUNT(*) AS publication_count,\n  SUM(AminerCitationCount) AS total_citations\nFROM AuthorPublications\nWHERE\n  -- Filter out any null or empty author names that might result from unnesting\n  author_name IS NOT NULL AND TRIM(author_name) <> ''\nGROUP BY\n  author_name\nORDER BY\n  total_citations DESC NULLS LAST, -- Prioritize authors with the most citations\n  publication_count DESC\nLIMIT 10;","fields":{"summary":"A ranked list of the top 10 most influential authors, measured by their total number of publications and the sum of their Aminer citation counts.","fields":[{"name":"author_name","label":"Author","description":"Derived from AuthorNames-Deduped: The author's name, obtained by unnesting the list of authors for each publication.","semantic_type":"Name","role":"primary"},{"name":"publication_count","label":"Publications","description":"Derived from the dataset records after unnesting AuthorNames-Deduped: The total count of publications, grouped by author_name.","semantic_type":"Numeric","role":"segment"},{"name":"total_citations","label":"Total Aminer Citations","description":"Derived from AminerCitationCount: The sum of citation counts from the Aminer database, grouped by author_name.","semantic_type":"Numeric","role":"segment"}]}},"Show the median PageLength (derived from LastPage - FirstPage) across different PaperType categories, to understand typical lengths for various publication formats.":{"goal":"Show the median PageLength (derived from LastPage - FirstPage) across different PaperType categories, to understand typical lengths for various publication formats.","sql":"-- Calculate the median page length for each paper type\nSELECT\n  COALESCE(NULLIF(TRIM(\"PaperType\"), ''), 'Unknown') AS paper_type,\n  CAST(MEDIAN(\"LastPage\" - \"FirstPage\" + 1) AS INTEGER) AS median_page_length\nFROM $tablename\nWHERE\n  \"LastPage\" IS NOT NULL\n  AND \"FirstPage\" IS NOT NULL\n  AND (\"LastPage\" - \"FirstPage\" + 1) > 0 -- Ensure page length is a positive value\nGROUP BY\n  1 -- group by paper_type\nORDER BY\n  2 DESC; -- order by median_page_length descending","fields":{"summary":"This view calculates the median page length for each publication type, providing insight into the typical length of different research formats.","fields":[{"name":"paper_type","label":"Paper Type","description":"Derived from PaperType: Cleaned by trimming whitespace and coalescing NULL or empty values into 'Unknown'.","semantic_type":"Category","role":"primary"},{"name":"median_page_length","label":"Median Page Length","description":"Derived from LastPage and FirstPage: Calculated as the median of (LastPage - FirstPage + 1) for each paper type, excluding records with invalid page numbers.","semantic_type":"Numeric","role":"primary"}]}},"Analyze the count of publications receiving Awards over Year, segmented by Conference, to observe trends in recognition and award distribution.":{"goal":"Analyze the count of publications receiving Awards over Year, segmented by Conference, to observe trends in recognition and award distribution.","sql":"-- Calculate the number of award-winning publications per year and conference\nSELECT\n  -- Convert the integer year to a DATE type for proper temporal analysis\n  MAKE_DATE(\"Year\", 1, 1) AS publication_year,\n  \"Conference\" AS conference,\n  COUNT(\"DOI\") AS award_winning_publications\nFROM $tablename\nWHERE\n  -- Filter for publications that have received at least one award\n  len(\"Award\") > 0\nGROUP BY\n  publication_year,\n  conference\nORDER BY\n  publication_year ASC,\n  conference ASC;","fields":{"summary":"A summary of the number of award-winning publications, aggregated by year and conference, to show trends in academic recognition.","fields":[{"name":"publication_year","label":"Publication Year","description":"Derived from Year: The publication year, converted to a DATE format representing January 1st of the given year.","semantic_type":"Year","role":"primary"},{"name":"conference","label":"Conference","description":"Original field Conference used without modification.","semantic_type":"Category","role":"segment"},{"name":"award_winning_publications","label":"Award-Winning Publications","description":"Derived from DOI and Award: A count of publications that have received at least one award (where the 'Award' list is not empty), grouped by publication year and conference.","semantic_type":"Numeric","role":"primary"}]}}}}