45:["$","title",null,{"children":"Positional Diction Clustering / Ordered Groups, now with a clearer input spec | glassmanlab | Observable"}] 4d:T2aed,function pdc_with_cleaner_input_spec( all_segments, absolute_sim_dist_threshold, percentage_same_response_threshold, group_making_threshold, a, b ) { //HELPERS //helper 1 -- definition of what's close; Defining a notion of distance with meta data (in the future, it may not be symmetric, especially if its specific to our visualization) let distance = (sentence_obj1, sentence_obj2) => { var mismatch_total = 0; var total_tokens = 0; var mismatch_dict = {}; var match_dict = {}; // could replace w a different distance measure, could be roughly BoW cosine // this could be based on greying for (const [key, value] of Object.entries(sentence_obj1.bow)) { let value2 = sentence_obj2.bow[key]; if (value2 === undefined) { mismatch_total += value; mismatch_dict[key] = value; total_tokens += value; } else { match_dict[key] = { value, value2 }; total_tokens += value + value2; } } for (const [key, value] of Object.entries(sentence_obj2.bow)) { if (sentence_obj1.bow[key] === undefined) { mismatch_total += value; mismatch_dict[key] = value; total_tokens += value; } } var mismatch_score = mismatch_total / total_tokens; return { mismatch_score, mismatch_total, total_tokens, mismatch_dict, match_dict }; }; //helper 2 --- compute pairwise distances let get_pairs = (minSimScoreThreshold, allSentences) => { let simMatrix = Array(allSentences.length) .fill(0) .map(() => Array(allSentences.length).fill(0)); // TODO: SAFEST THING IS TO FILL WITH INF NOT ZEROS let simMatrix_unmodified = Array(allSentences.length) .fill(0) .map(() => Array(allSentences.length).fill(0)); // TODO: SAFEST THING IS TO FILL WITH INF NOT ZEROS for (let i = 0; i < allSentences.length; i++) { for (let j = 0; j < i; j++) { //instead of j { let min = Infinity; let indices = []; for (let i = 0; i < arr.length; i++) { for (let j = 0; j < i; j++) { //TODO: EXPAND TO ARR.LENGTH IF ASSYMETRIC FUNCTION if (arr[i][j] < min) { min = arr[i][j]; indices = [i, j]; } } } return [min, indices]; }; let minSimScore; let bestPair; [minSimScore, bestPair] = getMaxSimIndices(simMatrix); clusters.push(bestPair); cluster_dists.push(minSimScore); simMatrix[bestPair[0]][bestPair[1]] = Infinity; while (minSimScore < minSimScoreThreshold) { [minSimScore, bestPair] = getMaxSimIndices(simMatrix); clusters.push(bestPair); cluster_dists.push(minSimScore); simMatrix[bestPair[0]][bestPair[1]] = Infinity; } // only output that is used is simMatrix_unmodified return { minSimScore, clusters, cluster_dists, simMatrix_unmodified }; }; //helper let getCurrentGroupIdx = (sentenceID, groups) => { for (let i = 0; i < groups.length; i++) { if (groups[i].includes(sentenceID)) { return i; } } return -1; }; //helper to check that there aren't too many segments in a cluster from the same document let getPercentageSharedResponses = (groupA, groupB, allSentences) => { let responses_included = []; for (let i = 0; i < groupA.length; i++) { responses_included.push(allSentences[groupA[i]].respIndex); } for (let i = 0; i < groupB.length; i++) { responses_included.push(allSentences[groupB[i]].respIndex); } let responses_included_unique = Array.from(new Set(responses_included)); return 1 - responses_included_unique.length / responses_included.length; }; let addToGroups = (groups, pair) => { let group_idx_pair0 = getCurrentGroupIdx(pair[0], groups); let group_idx_pair1 = getCurrentGroupIdx(pair[1], groups); if (group_idx_pair0 == -1 && group_idx_pair1 == -1) { groups.push(pair); } else if (group_idx_pair0 == group_idx_pair1) { return groups; //do nothing! } else if (group_idx_pair0 > -1 && group_idx_pair1 > -1) { //TODO: WHAT PERCENTAGE OF SENTENCES IN THIS CLUSTER SHARE A RESPONSE? IF 30% OR MORE SHARE THE SAME RESPONSE WHEN COMBINED, DON'T COMBINE if ( getPercentageSharedResponses( groups[group_idx_pair0], groups[group_idx_pair1], all_segments ) > percentage_same_response_threshold ) { return groups; } //WHICH IS BETTER FOR MINIMIZING DISTANCES OF ADJACENT GROUP MEMBERS---ADDING A TO B OR B TO A? let end_of_group_a = groups[group_idx_pair0].slice(-1); let start_of_group_b = groups[group_idx_pair1].slice(0, 1); let end_of_group_b = groups[group_idx_pair1].slice(-1); let start_of_group_a = groups[group_idx_pair0].slice(0, 1); let dist_a_push_b = simMatrix[end_of_group_a][start_of_group_b]; let dist_b_push_a = simMatrix[end_of_group_b][start_of_group_a]; if (dist_a_push_b <= dist_b_push_a) { let group_part_b = groups[group_idx_pair1]; for (let i = 0; i < group_part_b.length; i++) { groups[group_idx_pair0].push(group_part_b[i]); } groups.splice(group_idx_pair1, 1); //removes the group was added elsewhere } else { let group_part_a = groups[group_idx_pair0]; for (let i = 0; i < group_part_a.length; i++) { groups[group_idx_pair1].push(group_part_a[i]); } groups.splice(group_idx_pair0, 1); //removes the group was added elsewhere } //TODO: CONSIDER TESTING FOR MERGED CLUSTER PROPERTIES/STATS } else if (group_idx_pair0 > -1) { if ( getPercentageSharedResponses( groups[group_idx_pair0], [pair[1]], all_segments ) > percentage_same_response_threshold ) { return groups; } let start_of_group = groups[group_idx_pair0].slice(0, 1); let end_of_group = groups[group_idx_pair0].slice(-1); let dist_at_end_of_group = simMatrix[end_of_group][pair[1]]; let dist_at_start_of_group = simMatrix[pair[1]][start_of_group]; if (dist_at_start_of_group <= dist_at_end_of_group) { groups[group_idx_pair0].splice(0, 0, pair[1]); } else { groups[group_idx_pair0].splice( groups[group_idx_pair0].length, 0, pair[1] ); } //TODO: CONSIDER TESTING FOR MERGED CLUSTER PROPERTIES/STATS //TODO: WHAT PERCENTAGE OF SENTENCES IN THIS CLUSTER SHARE A RESPONSE? IF 30% OR MORE SHARE THE SAME RESPONSE WHEN COMBINED, DON'T COMBINE } else if (group_idx_pair1 > -1) { if ( getPercentageSharedResponses( groups[group_idx_pair1], [pair[0]], all_segments ) > percentage_same_response_threshold ) { return groups; } let start_of_group = groups[group_idx_pair1].slice(0, 1); let end_of_group = groups[group_idx_pair1].slice(-1); let dist_at_end_of_group = simMatrix[end_of_group][pair[0]]; let dist_at_start_of_group = simMatrix[pair[0]][start_of_group]; if (dist_at_start_of_group <= dist_at_end_of_group) { groups[group_idx_pair1].splice(0, 0, pair[0]); } else { groups[group_idx_pair1].splice( groups[group_idx_pair1].length, 0, pair[0] ); } //TODO: CONSIDER TESTING FOR MERGED CLUSTER PROPERTIES/STATS //TODO: WHAT PERCENTAGE OF SENTENCES IN THIS CLUSTER SHARE A RESPONSE? IF 30% OR MORE SHARE THE SAME RESPONSE WHEN COMBINED, DON'T COMBINE } return groups; //TODO: PUT IN LOGIC }; //MAIN let simMatrix = get_pairs( absolute_sim_dist_threshold, all_segments ).simMatrix_unmodified; let pairs = get_pairs(absolute_sim_dist_threshold, all_segments).clusters; //TODO: WHY IS GET_PAIRS RUN TWICE? // this computes the ordered clusters let groups = []; let ordered_pairs = get_pairs( absolute_sim_dist_threshold, all_segments ).clusters; for (let pair_idx = 0; pair_idx < ordered_pairs.length; pair_idx++) { let pair = ordered_pairs[pair_idx]; let pair_loc_abs_diff = Math.abs( all_segments[pair[0]].normalized_location - all_segments[pair[1]].normalized_location ); let pair_sim_dist = simMatrix[pair[0]][pair[1]]; // TODO: SAFEST IS TO CONSIDER BOTH DIRECTIONS OR MIN OF BOTH DIRECTIONS IF DISTANCE IS ASSYMMETRIC if (a * pair_sim_dist + b * pair_loc_abs_diff < group_making_threshold) { groups = addToGroups(groups, pair); //return groups } } // add singleton clusters to groups let flattened_groups = groups.flat(); for (let i = 0; i < all_segments.length; i++) { if (!flattened_groups.includes(i)) { groups.push([i]); // adds a singleton } } //return groups //now order it by average or median normalized location //compute aggregate normalized location for each group let ordered_groups = []; let median_groups_locations = []; let num_groups = groups.length; for (let group_idx = 0; group_idx < num_groups; group_idx++) { let group_locations = []; for ( let intra_group_idx = 0; intra_group_idx < groups[group_idx].length; intra_group_idx++ ) { group_locations.push( all_segments[groups[group_idx][intra_group_idx]].normalized_location ); } median_groups_locations.push(d3.mean(group_locations)); } //return median_groups_locations for (let i = 0; i < num_groups; i++) { let next_group_idx = median_groups_locations.indexOf( d3.min(median_groups_locations) ); //TODO: TIE BREAK TO PUT SENTENCES FROM LONGER DOCS FIRST //update: using mean fixed this ordered_groups.push(groups[next_group_idx]); groups.splice(next_group_idx, 1); median_groups_locations.splice(next_group_idx, 1); //return median_groups_locations } //for pretty printing let sentences_in_groups = []; for (let i = 0; i < ordered_groups.length; i++) { sentences_in_groups.push([]); for (let j = 0; j < ordered_groups[i].length; j++) { sentences_in_groups[i].push(all_segments[ordered_groups[i][j]].sentence); } } return sentences_in_groups; }4e:T3094,function pdc( segmented_documents, absolute_sim_dist_threshold, percentage_same_response_threshold, group_making_threshold, a, b ) { //HELPERS //helper 0 -- processing data let compute_response_lengths_and_add_normalized_locations = ( allSentences ) => { let response_lengths = {}; for (let i = 0; i < allSentences.length; i++) { if (Object.keys(response_lengths).includes(allSentences[i].respIndex)) { if ( response_lengths[allSentences[i].respIndex] < allSentences[i].sentIndex ) { response_lengths[allSentences[i].respIndex] = allSentences[i].sentIndex; } } else { response_lengths[allSentences[i].respIndex] = allSentences[i].sentIndex; } } for (let i = 0; i < allSentences.length; i++) { allSentences[i].total_resp_length = response_lengths[allSentences[i].respIndex]; if (allSentences[i].sentIndex == 0) { allSentences[i].normalized_location = 0; } else { allSentences[i].normalized_location = parseFloat(allSentences[i].sentIndex) / parseFloat(allSentences[i].total_resp_length); } } return response_lengths; }; //helper 1 -- definition of what's close; Defining a notion of distance with meta data (in the future, it may not be symmetric, especially if its specific to our visualization) let distance = (sentence_obj1, sentence_obj2) => { var mismatch_total = 0; var total_tokens = 0; var mismatch_dict = {}; var match_dict = {}; // could replace w a different distance measure, could be roughly BoW cosine // this could be based on greying for (const [key, value] of Object.entries(sentence_obj1.bow)) { let value2 = sentence_obj2.bow[key]; if (value2 === undefined) { mismatch_total += value; mismatch_dict[key] = value; total_tokens += value; } else { match_dict[key] = { value, value2 }; total_tokens += value + value2; } } for (const [key, value] of Object.entries(sentence_obj2.bow)) { if (sentence_obj1.bow[key] === undefined) { mismatch_total += value; mismatch_dict[key] = value; total_tokens += value; } } var mismatch_score = mismatch_total / total_tokens; return { mismatch_score, mismatch_total, total_tokens, mismatch_dict, match_dict }; }; //helper 2 --- compute pairwise distances let get_pairs = (minSimScoreThreshold, allSentences) => { let simMatrix = Array(allSentences.length) .fill(0) .map(() => Array(allSentences.length).fill(0)); // TODO: SAFEST THING IS TO FILL WITH INF NOT ZEROS let simMatrix_unmodified = Array(allSentences.length) .fill(0) .map(() => Array(allSentences.length).fill(0)); // TODO: SAFEST THING IS TO FILL WITH INF NOT ZEROS for (let i = 0; i < allSentences.length; i++) { for (let j = 0; j < i; j++) { //instead of j { let min = Infinity; let indices = []; for (let i = 0; i < arr.length; i++) { for (let j = 0; j < i; j++) { //TODO: EXPAND TO ARR.LENGTH IF ASSYMETRIC FUNCTION if (arr[i][j] < min) { min = arr[i][j]; indices = [i, j]; } } } return [min, indices]; }; let minSimScore; let bestPair; [minSimScore, bestPair] = getMaxSimIndices(simMatrix); clusters.push(bestPair); cluster_dists.push(minSimScore); simMatrix[bestPair[0]][bestPair[1]] = Infinity; while (minSimScore < minSimScoreThreshold) { [minSimScore, bestPair] = getMaxSimIndices(simMatrix); clusters.push(bestPair); cluster_dists.push(minSimScore); simMatrix[bestPair[0]][bestPair[1]] = Infinity; } // only output that is used is simMatrix_unmodified return { minSimScore, clusters, cluster_dists, simMatrix_unmodified }; }; //helper let getCurrentGroupIdx = (sentenceID, groups) => { for (let i = 0; i < groups.length; i++) { if (groups[i].includes(sentenceID)) { return i; } } return -1; }; //helper to check that there aren't too many segments in a cluster from the same document let getPercentageSharedResponses = (groupA, groupB, allSentences) => { let responses_included = []; for (let i = 0; i < groupA.length; i++) { responses_included.push(allSentences[groupA[i]].respIndex); } for (let i = 0; i < groupB.length; i++) { responses_included.push(allSentences[groupB[i]].respIndex); } let responses_included_unique = Array.from(new Set(responses_included)); return 1 - responses_included_unique.length / responses_included.length; }; let addToGroups = (groups,pair) => { let group_idx_pair0 = getCurrentGroupIdx(pair[0], groups) let group_idx_pair1 = getCurrentGroupIdx(pair[1], groups) if (group_idx_pair0 == -1 && group_idx_pair1 == -1) { groups.push(pair); } else if (group_idx_pair0 == group_idx_pair1){ return groups //do nothing! } else if (group_idx_pair0 > -1 && group_idx_pair1 > -1) { //TODO: WHAT PERCENTAGE OF SENTENCES IN THIS CLUSTER SHARE A RESPONSE? IF 30% OR MORE SHARE THE SAME RESPONSE WHEN COMBINED, DON'T COMBINE if (getPercentageSharedResponses(groups[group_idx_pair0],groups[group_idx_pair1],all_segments) > percentage_same_response_threshold) { return groups } //WHICH IS BETTER FOR MINIMIZING DISTANCES OF ADJACENT GROUP MEMBERS---ADDING A TO B OR B TO A? let end_of_group_a = groups[group_idx_pair0].slice(-1); let start_of_group_b = groups[group_idx_pair1].slice(0,1); let end_of_group_b = groups[group_idx_pair1].slice(-1); let start_of_group_a = groups[group_idx_pair0].slice(0,1); let dist_a_push_b = simMatrix[end_of_group_a][start_of_group_b]; let dist_b_push_a = simMatrix[end_of_group_b][start_of_group_a]; if (dist_a_push_b <= dist_b_push_a) { let group_part_b = groups[group_idx_pair1]; for (let i = 0; i -1){ if (getPercentageSharedResponses(groups[group_idx_pair0],[pair[1]],all_segments) > percentage_same_response_threshold) { return groups } let start_of_group = groups[group_idx_pair0].slice(0,1); let end_of_group = groups[group_idx_pair0].slice(-1); let dist_at_end_of_group = simMatrix[end_of_group][pair[1]]; let dist_at_start_of_group = simMatrix[pair[1]][start_of_group]; if (dist_at_start_of_group <= dist_at_end_of_group){ groups[group_idx_pair0].splice(0,0,pair[1]); } else{ groups[group_idx_pair0].splice(groups[group_idx_pair0].length,0,pair[1]); } //TODO: CONSIDER TESTING FOR MERGED CLUSTER PROPERTIES/STATS //TODO: WHAT PERCENTAGE OF SENTENCES IN THIS CLUSTER SHARE A RESPONSE? IF 30% OR MORE SHARE THE SAME RESPONSE WHEN COMBINED, DON'T COMBINE } else if (group_idx_pair1 > -1){ if (getPercentageSharedResponses(groups[group_idx_pair1],[pair[0]],all_segments) > percentage_same_response_threshold) { return groups } let start_of_group = groups[group_idx_pair1].slice(0,1); let end_of_group = groups[group_idx_pair1].slice(-1); let dist_at_end_of_group = simMatrix[end_of_group][pair[0]]; let dist_at_start_of_group = simMatrix[pair[0]][start_of_group]; if (dist_at_start_of_group <= dist_at_end_of_group){ groups[group_idx_pair1].splice(0,0,pair[0]); } else { groups[group_idx_pair1].splice(groups[group_idx_pair1].length,0,pair[0]); } //TODO: CONSIDER TESTING FOR MERGED CLUSTER PROPERTIES/STATS //TODO: WHAT PERCENTAGE OF SENTENCES IN THIS CLUSTER SHARE A RESPONSE? IF 30% OR MORE SHARE THE SAME RESPONSE WHEN COMBINED, DON'T COMBINE } return groups; //TODO: PUT IN LOGIC } //MAIN //extract and format relevant data from segmented_documents let all_segments = segmented_documents .map((respObj) => { return respObj.sentences.map((sentObj, idx) => { const sentenceDetails = { bow: sentObj.bow, respIndex: respObj.index, sentIndex: idx, sentence: sentObj.text }; return sentenceDetails; }); }) .flat(); //add normalized locations of segments within documents; changes all_segments in place let response_lengths = compute_response_lengths_and_add_normalized_locations(all_segments); let simMatrix = get_pairs( absolute_sim_dist_threshold, all_segments ).simMatrix_unmodified; let pairs = get_pairs(absolute_sim_dist_threshold, all_segments).clusters; //TODO: WHY IS GET_PAIRS RUN TWICE? // this computes the ordered clusters let groups = []; let ordered_pairs = get_pairs(absolute_sim_dist_threshold, all_segments).clusters; for (let pair_idx = 0; pair_idx < ordered_pairs.length; pair_idx++) { let pair = ordered_pairs[pair_idx]; let pair_loc_abs_diff = Math.abs( all_segments[pair[0]].normalized_location - all_segments[pair[1]].normalized_location ); let pair_sim_dist = simMatrix[pair[0]][pair[1]]; // TODO: SAFEST IS TO CONSIDER BOTH DIRECTIONS OR MIN OF BOTH DIRECTIONS IF DISTANCE IS ASSYMMETRIC if (a * pair_sim_dist + b * pair_loc_abs_diff < group_making_threshold) { groups = addToGroups(groups, pair); //return groups } } // add singleton clusters to groups let flattened_groups = groups.flat(); for (let i=0; i {\n return respObj.sentences.map((sentObj, idx) => {\n const sentenceDetails = {\n bow: sentObj.bow,\n respIndex: respObj.index,\n sentIndex: idx,\n sentence: sentObj.text\n };\n return sentenceDetails;\n });\n })\n .flat()","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":150,"value":"### Expected output: PDC on LLM responses","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"","database":"$undefined"},{"id":142,"value":"jsonPDCofLLMResponses = FileAttachment(\"sentences_in_groups.json\").json()","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":157,"value":"### Testing for equivalence of algorithm output and saved output","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"","database":"$undefined"},{"id":334,"value":"deepEqualsListOfLists(\n jsonPDCofLLMResponses,\n pdc_with_cleaner_input_spec(cleanedJSONLLMResponses, 0.55, 0.1, 1.2, 1.5, 1)\n)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":173,"value":"deepEqualsListOfLists(jsonPDCofLLMResponses, pdc(jsonLLMResponses,0.55,\n 0.1,\n 1.2,\n 1.5,\n 1))","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":248,"value":"### Confirming output equivalence via visual inspection","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"","database":"$undefined"},{"id":332,"value":"pdc_with_cleaner_input_spec(cleanedJSONLLMResponses, 0.55, 0.1, 1.2, 1.5, 1)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":234,"value":"pdc(jsonLLMResponses,0.55,\n 0.1,\n 1.2,\n 1.5,\n 1)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":253,"value":"# Default parameter values","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"","database":"$undefined"},{"id":27,"value":"absolute_sim_dist_threshold = 0.55 //absolute_sim_dist_threshold is the maximum distance between two segments in order to be considered close enough to be clustered together.","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":100,"value":"percentage_same_response_threshold = 0.1 //The maximum fraction of segments from the same documents in a group allowed","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":48,"value":"group_making_threshold = 1.2; // For a given pair of segments, the linear mixture of the absolute difference in normalized segment location and content similarity (defined by b and a respectively) must be below the group_making_threshold for the pair to be considered for inclusion in the same group.","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":51,"value":"a = 1.5; //relative weight on content similarity","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":53,"value":"b = 1; //relative weight on location coherence (less important than content, I think)","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":328,"value":"# PDC code with cleaner expected input format","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"","database":"$undefined"},{"id":330,"value":"$4d","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":250,"value":"# Original PDC Code (difficult input format)","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"","database":"$undefined"},{"id":4,"value":"d3 = require(\"d3\")","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":199,"value":"$4e","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W4f","stdlib":"1"}} 44:null