Serena branch for metadata from ArXiv

data/.ipynb_checkpoints/sparql queries-checkpoint.ipynb

@@ -3,7 +3,7 @@
   {
    "cell_type": "code",
    "execution_count": 13,
-   "id": "spanish-group",
+   "id": "alive-divorce",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -12,8 +12,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 35,
-   "id": "cooked-heating",
+   "execution_count": 45,
+   "id": "split-wagner",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -23,8 +23,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 36,
-   "id": "atmospheric-feeding",
+   "execution_count": 46,
+   "id": "federal-trustee",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -34,8 +34,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 34,
-   "id": "confidential-phenomenon",
+   "execution_count": 47,
+   "id": "latter-apparatus",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -48,7 +48,7 @@
     "\n",
     "query2 = \"\"\"\n",
     "SELECT DISTINCT ?uniName WHERE { \n",
-    "  ?uni owl:sameAs ?wikidataEntity\n",
+    "  ?uni owl:sameAs ?wikidataEntity .\n",
     "  ?uni ccso:legalName ?uniName\n",
     "}\n",
     "\"\"\"\n",
@@ -64,13 +64,19 @@
     " ?uni ccso:legalName ?uniName .\n",
     " ?uni owl:sameAs ?wikidataEntity\n",
     "}\n",
-    "\"\"\""
+    "\"\"\"\n",
+    "\n",
+    "query5 = \"\"\"\n",
+    "SELECT DISTINCT ?uniName WHERE {\n",
+    " ?deUniversity ccso:legalName ?uniName .\n",
+    " ?deUniversity owl:sameAs ?wikidataEntity \n",
+    "}\"\"\""
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 19,
-   "id": "chicken-crossing",
+   "id": "chinese-surname",
    "metadata": {},
    "outputs": [
     {
@@ -92,29 +98,29 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
-   "id": "abstract-tobacco",
+   "execution_count": 43,
+   "id": "worth-sensitivity",
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "<Graph identifier=N80d3e852c0524ee7bc23253702bfb729 (<class 'rdflib.graph.Graph'>)>"
+       "<rdflib.plugins.sparql.processor.SPARQLResult at 0x23a81f267c0>"
       ]
      },
-     "execution_count": 9,
+     "execution_count": 43,
      "metadata": {},
      "output_type": "execute_result"
     }
    ],
    "source": [
-    "g"
+    "g.query(query2)"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 29,
-   "id": "attempted-people",
+   "id": "young-store",
    "metadata": {},
    "outputs": [
     {
@@ -135,17 +141,17 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 37,
-   "id": "characteristic-round",
+   "execution_count": 48,
+   "id": "chief-command",
    "metadata": {},
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "<rdflib.plugins.sparql.processor.SPARQLResult at 0x23a81ba0280>"
+       "<rdflib.plugins.sparql.processor.SPARQLResult at 0x23a81fe5730>"
       ]
      },
-     "execution_count": 37,
+     "execution_count": 48,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -157,17 +163,45 @@
   {
    "cell_type": "code",
    "execution_count": 12,
-   "id": "wired-atlantic",
+   "id": "coated-politics",
    "metadata": {},
    "outputs": [],
    "source": [
     "g.serialize(destination='q1_output.txt', format='turtle')"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": 49,
+   "id": "behavioral-america",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "g.serialize(destination='q4_output.txt', format='turtle')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 44,
+   "id": "amazing-labor",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "g.serialize(destination='q2_output.txt', format='turtle')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "colonial-midnight",
+   "metadata": {},
+   "source": [
+    "https://www.w3.org/TR/sparql11-query/"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "imported-spectrum",
+   "id": "russian-threshold",
    "metadata": {},
    "outputs": [],
    "source": []

data/.ipynb_checkpoints/text_manipulation-checkpoint.ipynb

@@ -0,0 +1,44 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "authorized-banks",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# tab -> space\n",
+    "\n",
+    "inputFile = open(“65.texts.txt”, “r”) \n",
+    "exportFile = open(“65texts.txt”, “w”)\n",
+    "for line in inputFile:\n",
+    "   new_line = line.replace('\\t', ' ')\n",
+    "   exportFile.write(new_line) \n",
+    "\n",
+    "inputFile.close()\n",
+    "exportFile.close()"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.9.1"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}