Datasets:

UCSD26
/

medical_dialog

- {"en": {"description": "The MedDialog dataset (English) contains conversations (in English) between doctors and patients.It has 0.26 million dialogues. The data is continuously growing and more dialogues will be added. The raw dialogues are from healthcaremagic.com and icliniq.com.\nAll copyrights of the data belong to healthcaremagic.com and icliniq.com.\n", "citation": "@article{chen2020meddiag,\n title={MedDialog: a large-scale medical dialogue dataset},\n author={Chen, Shu and Ju, Zeqian and Dong, Xiangyu and Fang, Hongchao and Wang, Sicheng and Yang, Yue and Zeng, Jiaqi and Zhang, Ruisi and Zhang, Ruoyu and Zhou, Meng and Zhu, Penghui and Xie, Pengtao},\n journal={arXiv preprint arXiv:2004.03329},\n year={2020}\n}\n", "homepage": "https://github.com/UCSD-AI4H/Medical-Dialogue-System", "license": "", "features": {"file_name": {"dtype": "string", "id": null, "_type": "Value"}, "dialogue_id": {"dtype": "int32", "id": null, "_type": "Value"}, "dialogue_url": {"dtype": "string", "id": null, "_type": "Value"}, "dialogue_turns": {"feature": {"speaker": {"num_classes": 2, "names": ["Patient", "Doctor"], "id": null, "_type": "ClassLabel"}, "utterance": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "medical_dialog", "config_name": "en", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 290274759, "num_examples": 229674, "dataset_name": "medical_dialog"}}, "download_checksums": {}, "download_size": 0, "post_processing_size": null, "dataset_size": 290274759, "size_in_bytes": 290274759}, "zh": {"description": "The MedDialog dataset (English) contains conversations (in English) between doctors and patients.It has 0.26 million dialogues. The data is continuously growing and more dialogues will be added. The raw dialogues are from healthcaremagic.com and icliniq.com.\nAll copyrights of the data belong to healthcaremagic.com and icliniq.com.\n", "citation": "@article{chen2020meddiag,\n title={MedDialog: a large-scale medical dialogue dataset},\n author={Chen, Shu and Ju, Zeqian and Dong, Xiangyu and Fang, Hongchao and Wang, Sicheng and Yang, Yue and Zeng, Jiaqi and Zhang, Ruisi and Zhang, Ruoyu and Zhou, Meng and Zhu, Penghui and Xie, Pengtao},\n journal={arXiv preprint arXiv:2004.03329},\n year={2020}\n}\n", "homepage": "https://github.com/UCSD-AI4H/Medical-Dialogue-System", "license": "", "features": {"file_name": {"dtype": "string", "id": null, "_type": "Value"}, "dialogue_id": {"dtype": "int32", "id": null, "_type": "Value"}, "dialogue_url": {"dtype": "string", "id": null, "_type": "Value"}, "dialogue_turns": {"feature": {"speaker": {"num_classes": 2, "names": ["\u75c5\u4eba", "\u533b\u751f"], "id": null, "_type": "ClassLabel"}, "utterance": {"dtype": "string", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "medical_dialog", "config_name": "zh", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1092063621, "num_examples": 1921127, "dataset_name": "medical_dialog"}}, "download_checksums": {}, "download_size": 0, "post_processing_size": null, "dataset_size": 1092063621, "size_in_bytes": 1092063621}, "processed.en": {"description": "The MedDialog dataset (English) contains conversations (in English) between doctors and patients.It has 0.26 million dialogues. The data is continuously growing and more dialogues will be added. The raw dialogues are from healthcaremagic.com and icliniq.com.\nAll copyrights of the data belong to healthcaremagic.com and icliniq.com.\n", "citation": "@article{chen2020meddiag,\n title={MedDialog: a large-scale medical dialogue dataset},\n author={Chen, Shu and Ju, Zeqian and Dong, Xiangyu and Fang, Hongchao and Wang, Sicheng and Yang, Yue and Zeng, Jiaqi and Zhang, Ruisi and Zhang, Ruoyu and Zhou, Meng and Zhu, Penghui and Xie, Pengtao},\n journal={arXiv preprint arXiv:2004.03329},\n year={2020}\n}\n", "homepage": "https://github.com/UCSD-AI4H/Medical-Dialogue-System", "license": "Copyright", "features": {"description": {"dtype": "string", "id": null, "_type": "Value"}, "utterances": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "medical_dialog", "config_name": "processed.en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 370745, "num_examples": 482, "dataset_name": "medical_dialog"}, "validation": {"name": "validation", "num_bytes": 52145, "num_examples": 60, "dataset_name": "medical_dialog"}, "test": {"name": "test", "num_bytes": 46514, "num_examples": 61, "dataset_name": "medical_dialog"}}, "download_checksums": {"https://drive.google.com/uc?export=download&id=1ria4E6IdTIPsikL4Glm3uy1tFKJKw0W8": {"num_bytes": 414490, "checksum": "568a9c6c670502eec3319c78e9d12c0aebb883c0d1e45095b5dd5f99d8b6b874"}, "https://drive.google.com/uc?export=download&id=1KAZneuwdfEVQQM6euCX4pMDP-9DQpiB5": {"num_bytes": 57706, "checksum": "a5cd29f17fcfedf01af41410e12e47474ba1176f376136e18fc0446b7e2f52b2"}, "https://drive.google.com/uc?export=download&id=10izqL71kcgnteYsf87Vh6j_mZ8sZM2Rc": {"num_bytes": 52018, "checksum": "316e5b3eb03ec7210b0d84414df0e84a42b396205d72a2b5fdba533fd19a5ebd"}}, "download_size": 524214, "post_processing_size": null, "dataset_size": 469404, "size_in_bytes": 993618}, "processed.zh": {"description": "The MedDialog dataset (English) contains conversations (in English) between doctors and patients.It has 0.26 million dialogues. The data is continuously growing and more dialogues will be added. The raw dialogues are from healthcaremagic.com and icliniq.com.\nAll copyrights of the data belong to healthcaremagic.com and icliniq.com.\n", "citation": "@article{chen2020meddiag,\n title={MedDialog: a large-scale medical dialogue dataset},\n author={Chen, Shu and Ju, Zeqian and Dong, Xiangyu and Fang, Hongchao and Wang, Sicheng and Yang, Yue and Zeng, Jiaqi and Zhang, Ruisi and Zhang, Ruoyu and Zhou, Meng and Zhu, Penghui and Xie, Pengtao},\n journal={arXiv preprint arXiv:2004.03329},\n year={2020}\n}\n", "homepage": "https://github.com/UCSD-AI4H/Medical-Dialogue-System", "license": "Copyright", "features": {"utterances": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "medical_dialog", "config_name": "processed.zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1571262099, "num_examples": 2725989, "dataset_name": "medical_dialog"}, "validation": {"name": "validation", "num_bytes": 197117565, "num_examples": 340748, "dataset_name": "medical_dialog"}, "test": {"name": "test", "num_bytes": 196526738, "num_examples": 340754, "dataset_name": "medical_dialog"}}, "download_checksums": {"https://drive.google.com/uc?export=download&id=1AaDJoHaiHAwEZwtskRH8oL1UP4FRgmgx": {"num_bytes": 1665206303, "checksum": "fd34385487755d95783cf834921bff14ceb74d9a244962577140c9e291dce4e9"}, "https://drive.google.com/uc?export=download&id=1TvfZCmQqP1kURIfEinOcj5VOPelTuGwI": {"num_bytes": 208871784, "checksum": "ed6b04ff4d62a4fa5b5b85327d692302b3369c0d28e9da887c12ec78ea778ce4"}, "https://drive.google.com/uc?export=download&id=1pmmG95Yl6mMXRXDDSRb9-bYTxOE7ank5": {"num_bytes": 208276068, "checksum": "b1118b614f866089a1daf18107a72dd5ba77c50a1e9ca145491ddcef89d797b7"}}, "download_size": 2082354155, "post_processing_size": null, "dataset_size": 1964906402, "size_in_bytes": 4047260557}}

medical_dialog.py CHANGED Viewed

@@ -46,15 +46,21 @@ _LICENSE = "Unknown"
 # URLS of processed data
 _URLS = {
-    "en": {
-        "train": "https://drive.google.com/uc?export=download&id=1ria4E6IdTIPsikL4Glm3uy1tFKJKw0W8",
-        "validation": "https://drive.google.com/uc?export=download&id=1KAZneuwdfEVQQM6euCX4pMDP-9DQpiB5",
-        "test": "https://drive.google.com/uc?export=download&id=10izqL71kcgnteYsf87Vh6j_mZ8sZM2Rc",
     },
-    "zh": {
-        "train": "https://drive.google.com/uc?export=download&id=1AaDJoHaiHAwEZwtskRH8oL1UP4FRgmgx",
-        "validation": "https://drive.google.com/uc?export=download&id=1TvfZCmQqP1kURIfEinOcj5VOPelTuGwI",
-        "test": "https://drive.google.com/uc?export=download&id=1pmmG95Yl6mMXRXDDSRb9-bYTxOE7ank5",
     },
 }
@@ -77,33 +83,6 @@ class MedicalDialog(datasets.GeneratorBasedBuilder):
         ),
     ]
-    @property
-    def manual_download_instructions(self):
-        *processed, _ = self.config.name.split(".")
-        return (
-            None
-            if processed
-            else """\
-    \n  For English:\nYou need to go to https://drive.google.com/drive/folders/1g29ssimdZ6JzTST6Y8g6h-ogUNReBtJD?usp=sharing,\
-    and manually download the dataset from Google Drive. Once it is completed,
-    a file named Medical-Dialogue-Dataset-English-<timestamp-info>.zip will appear in your Downloads folder(
-    or whichever folder your browser chooses to save files to). Unzip the folder to obtain
-    a folder named "Medical-Dialogue-Dataset-English" several text files.
-    Now, you can specify the path to this folder for the data_dir argument in the
-    datasets.load_dataset(...) option.
-    The <path/to/folder> can e.g. be "/Downloads/Medical-Dialogue-Dataset-English".
-    The data can then be loaded using the below command:\
-         `datasets.load_dataset("medical_dialog", name="en", data_dir="/Downloads/Medical-Dialogue-Dataset-English")`.
-    \n  For Chinese:\nFollow the above process. Change the 'name' to 'zh'.The download link is https://drive.google.com/drive/folders/1r09_i8nJ9c1nliXVGXwSqRYqklcHd9e2
-    **NOTE**
-    - A caution while downloading from drive. It is better to download single files since creating a zip might not include files <500 MB. This has been observed mutiple times.
-    - After downloading the files and adding them to the appropriate folder, the path of the folder can be given as input tu the data_dir path.
-    """
-        )
     def _info(self):
         if self.config.name == "zh":
             features = datasets.Features(
@@ -158,23 +137,13 @@ class MedicalDialog(datasets.GeneratorBasedBuilder):
         """Returns SplitGenerators."""
         *processed, lang = self.config.name.split(".")
         if processed:
-            data_dir = dl_manager.download(_URLS[lang])
             splits = [datasets.Split.TRAIN, datasets.Split.VALIDATION, datasets.Split.TEST]
-            return [datasets.SplitGenerator(name=split, gen_kwargs={"filepaths": data_dir[split]}) for split in splits]
         else:
-            path_to_manual_file = os.path.abspath(os.path.expanduser(dl_manager.manual_dir))
-            if not os.path.exists(path_to_manual_file):
-                raise FileNotFoundError(
-                    f"{path_to_manual_file} does not exist. Make sure you insert a manual dir via `datasets.load_dataset('medical_dialog', data_dir=...)`. Manual download instructions: {self.manual_download_instructions})"
-                )
-            filepaths = [
-                os.path.join(path_to_manual_file, txt_file_name)
-                for txt_file_name in sorted(os.listdir(path_to_manual_file))
-                if txt_file_name.endswith("txt")
-            ]
-            return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": filepaths})]
     def _generate_examples(self, filepaths):
         """Yields examples. Iterates over each file and give the creates the corresponding features.
@@ -205,130 +174,130 @@ class MedicalDialog(datasets.GeneratorBasedBuilder):
                                 array = ""
         else:
             id_ = -1
-            for filepath in filepaths:
-                with open(filepath, encoding="utf-8") as f_in:
-                    # Parameters to just "sectionize" the raw data
-                    last_part = ""
-                    last_dialog = {}
-                    last_list = []
-                    last_user = ""
-                    check_list = []
-                    # These flags are present to have a single function address both chinese and english data
-                    # English data is a little hahazard (i.e. the sentences spans multiple different lines),
-                    # Chinese is compact with one line for doctor and patient.
-                    conv_flag = False
-                    des_flag = False
-                    while True:
-                        line = f_in.readline()
-                        if not line:
-                            break
-                        # Extracting the dialog id
-                        if line[:2] == "id":  # Hardcode alert!
-                            # Handling ID references that may come in the description
-                            # These were observed in the Chinese dataset and were not
-                            # followed by numbers
-                            try:
-                                dialogue_id = int(re.findall(r"\d+", line)[0])
-                            except IndexError:
-                                continue
-                        # Extracting the url
-                        if line[:4] == "http":  # Hardcode alert!
-                            dialogue_url = line.rstrip()
-                        # Extracting the patient info from description.
-                        if line[:11] == "Description":  # Hardcode alert!
-                            last_part = "description"
-                            last_dialog = {}
-                            last_list = []
-                            last_user = ""
-                            last_conv = {"speaker": "", "utterance": ""}
                             while True:
-                                line = f_in.readline()
                                 if (not line) or (line in ["\n", "\n\r"]):
                                     break
-                                else:
-                                    if data_lang == "zh":  # Condition in chinese
-                                        if line[:5] == "病情描述：":  # Hardcode alert!
-                                            last_user = "病人"
-                                            sen = f_in.readline().rstrip()
-                                            des_flag = True
-                                    if data_lang == "en":
-                                        last_user = "Patient"
-                                        sen = line.rstrip()
-                                        des_flag = True
-                                    if des_flag:
-                                        if sen == "":
-                                            continue
-                                        if sen in check_list:
-                                            last_conv["speaker"] = ""
-                                            last_conv["utterance"] = ""
-                                        else:
-                                            last_conv["speaker"] = last_user
-                                            last_conv["utterance"] = sen
-                                            check_list.append(sen)
-                                        des_flag = False
-                                        break
-                        # Extracting the conversation info from dialogue.
-                        elif line[:8] == "Dialogue":  # Hardcode alert!
-                            if last_part == "description" and len(last_conv["utterance"]) > 0:
-                                last_part = "dialogue"
                                 if data_lang == "zh":
-                                    last_user = "病人"
                                 if data_lang == "en":
-                                    last_user = "Patient"
-                                while True:
-                                    line = f_in.readline()
-                                    if (not line) or (line in ["\n", "\n\r"]):
-                                        conv_flag = False
-                                        last_user = ""
                                         last_list.append(copy.deepcopy(last_conv))
-                                        # To ensure close of conversation, only even number of sentences
-                                        # are extracted
-                                        last_turn = len(last_list)
-                                        if int(last_turn / 2) > 0:
-                                            temp = int(last_turn / 2)
-                                            id_ += 1
-                                            last_dialog["file_name"] = filepath
-                                            last_dialog["dialogue_id"] = dialogue_id
-                                            last_dialog["dialogue_url"] = dialogue_url
-                                            last_dialog["dialogue_turns"] = last_list[: temp * 2]
-                                            yield id_, last_dialog
-                                        break
-                                    if data_lang == "zh":
-                                        if line[:3] == "病人：" or line[:3] == "医生：":  # Hardcode alert!
-                                            user = line[:2]  # Hardcode alert!
-                                            line = f_in.readline()
-                                            conv_flag = True
-                                    # The elif block is to ensure that multi-line sentences are captured.
-                                    # This has been observed only in english.
-                                    if data_lang == "en":
-                                        if line.strip() == "Patient:" or line.strip() == "Doctor:":  # Hardcode alert!
-                                            user = line.replace(":", "").rstrip()
-                                            line = f_in.readline()
-                                            conv_flag = True
-                                        elif line[:2] != "id":  # Hardcode alert!
-                                            conv_flag = True
-                                    # Continues till the next ID is parsed
-                                    if conv_flag:
-                                        sen = line.rstrip()
-                                        if sen == "":
-                                            continue
-                                        if user == last_user:
-                                            last_conv["utterance"] = last_conv["utterance"] + sen
-                                        else:
-                                            last_user = user
-                                            last_list.append(copy.deepcopy(last_conv))
-                                            last_conv["utterance"] = sen
-                                            last_conv["speaker"] = user

 # URLS of processed data
 _URLS = {
+    "en": "data/Medical-Dialogue-Dataset-English.zip",
+    "zh": "data/Medical-Dialogue-Dataset-Chinese.zip",
+    "processed.en": "data/processed-english.zip",
+    "processed.zh": "data/processed-chinese.zip",
+}
+_FILENAMES = {
+    "processed.en": {
+        "train": "english-train.json",
+        "validation": "english-dev.json",
+        "test": "english-test.json",
     },
+    "processed.zh": {
+        "train": "train_data.json",
+        "validation": "validate_data.json",
+        "test": "test_data.json",
     },
 }
         ),
     ]
     def _info(self):
         if self.config.name == "zh":
             features = datasets.Features(
         """Returns SplitGenerators."""
         *processed, lang = self.config.name.split(".")
         if processed:
+            # data_dir = dl_manager.download(_URLS[lang])
+            data_dir = dl_manager.download_and_extract(_URLS[self.config.name])
             splits = [datasets.Split.TRAIN, datasets.Split.VALIDATION, datasets.Split.TEST]
+            return [datasets.SplitGenerator(name=split, gen_kwargs={"filepaths": os.path.join(data_dir, _FILENAMES[self.config.name][split])}) for split in splits]
         else:
+            archive = dl_manager.download(_URLS[self.config.name])
+            return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": dl_manager.iter_archive(archive)})]
     def _generate_examples(self, filepaths):
         """Yields examples. Iterates over each file and give the creates the corresponding features.
                                 array = ""
         else:
             id_ = -1
+            for filepath, f_in in filepaths:
+                # with open(filepath, encoding="utf-8") as f_in:
+                # Parameters to just "sectionize" the raw data
+                last_part = ""
+                last_dialog = {}
+                last_list = []
+                last_user = ""
+                check_list = []
+                # These flags are present to have a single function address both chinese and english data
+                # English data is a little hahazard (i.e. the sentences spans multiple different lines),
+                # Chinese is compact with one line for doctor and patient.
+                conv_flag = False
+                des_flag = False
+                while True:
+                    line = f_in.readline().decode("utf-8")
+                    if not line:
+                        break
+                    # Extracting the dialog id
+                    if line[:2] == "id":  # Hardcode alert!
+                        # Handling ID references that may come in the description
+                        # These were observed in the Chinese dataset and were not
+                        # followed by numbers
+                        try:
+                            dialogue_id = int(re.findall(r"\d+", line)[0])
+                        except IndexError:
+                            continue
+                    # Extracting the url
+                    if line[:4] == "http":  # Hardcode alert!
+                        dialogue_url = line.rstrip()
+                    # Extracting the patient info from description.
+                    if line[:11] == "Description":  # Hardcode alert!
+                        last_part = "description"
+                        last_dialog = {}
+                        last_list = []
+                        last_user = ""
+                        last_conv = {"speaker": "", "utterance": ""}
+                        while True:
+                            line = f_in.readline().decode("utf-8")
+                            if (not line) or (line in ["\n", "\n\r"]):
+                                break
+                            else:
+                                if data_lang == "zh":  # Condition in chinese
+                                    if line[:5] == "病情描述：":  # Hardcode alert!
+                                        last_user = "病人"
+                                        sen = f_in.readline().decode("utf-8").rstrip()
+                                        des_flag = True
+                                if data_lang == "en":
+                                    last_user = "Patient"
+                                    sen = line.rstrip()
+                                    des_flag = True
+                                if des_flag:
+                                    if sen == "":
+                                        continue
+                                    if sen in check_list:
+                                        last_conv["speaker"] = ""
+                                        last_conv["utterance"] = ""
+                                    else:
+                                        last_conv["speaker"] = last_user
+                                        last_conv["utterance"] = sen
+                                        check_list.append(sen)
+                                    des_flag = False
+                                    break
+                    # Extracting the conversation info from dialogue.
+                    elif line[:8] == "Dialogue":  # Hardcode alert!
+                        if last_part == "description" and len(last_conv["utterance"]) > 0:
+                            last_part = "dialogue"
+                            if data_lang == "zh":
+                                last_user = "病人"
+                            if data_lang == "en":
+                                last_user = "Patient"
                             while True:
+                                line = f_in.readline().decode("utf-8")
                                 if (not line) or (line in ["\n", "\n\r"]):
+                                    conv_flag = False
+                                    last_user = ""
+                                    last_list.append(copy.deepcopy(last_conv))
+                                    # To ensure close of conversation, only even number of sentences
+                                    # are extracted
+                                    last_turn = len(last_list)
+                                    if int(last_turn / 2) > 0:
+                                        temp = int(last_turn / 2)
+                                        id_ += 1
+                                        last_dialog["file_name"] = filepath
+                                        last_dialog["dialogue_id"] = dialogue_id
+                                        last_dialog["dialogue_url"] = dialogue_url
+                                        last_dialog["dialogue_turns"] = last_list[: temp * 2]
+                                        yield id_, last_dialog
                                     break
                                 if data_lang == "zh":
+                                    if line[:3] == "病人：" or line[:3] == "医生：":  # Hardcode alert!
+                                        user = line[:2]  # Hardcode alert!
+                                        line = f_in.readline().decode("utf-8")
+                                        conv_flag = True
+                                # The elif block is to ensure that multi-line sentences are captured.
+                                # This has been observed only in english.
                                 if data_lang == "en":
+                                    if line.strip() == "Patient:" or line.strip() == "Doctor:":  # Hardcode alert!
+                                        user = line.replace(":", "").rstrip()
+                                        line = f_in.readline().decode("utf-8")
+                                        conv_flag = True
+                                    elif line[:2] != "id":  # Hardcode alert!
+                                        conv_flag = True
+                                # Continues till the next ID is parsed
+                                if conv_flag:
+                                    sen = line.rstrip()
+                                    if sen == "":
+                                        continue
+                                    if user == last_user:
+                                        last_conv["utterance"] = last_conv["utterance"] + sen
+                                    else:
+                                        last_user = user
                                         last_list.append(copy.deepcopy(last_conv))
+                                        last_conv["utterance"] = sen
+                                        last_conv["speaker"] = user